fix(ai): preserve websocket upgrade diagnostics

Co-authored-by: nexxeln <95541290+nexxeln@users.noreply.github.com>

nix/hashes.json

@@ -1,8 +1,8 @@
 {
   "nodeModules": {
-    "x86_64-linux": "sha256-9IJxoe/MdL6nmoeKvxZop+77HJ/b3HbmpytNUvLqYkc=",
-    "aarch64-linux": "sha256-KR1J102RDTRDZIkAD3jQfXeP1G2DN9Es7KkdKo+daec=",
-    "aarch64-darwin": "sha256-HCgSoq1W6XU6m73Ck3wV8gjB687caUTM3w/EO+V7wsE=",
-    "x86_64-darwin": "sha256-GrKTDvjg0XeDIWbOvayWQjj0SdJd8WPm2+q3IVS17Jg="
+    "x86_64-linux": "sha256-JEqi00PCle+o5OfBlJJaZtXd+4sYB3o+rvYiESlN4dY=",
+    "aarch64-linux": "sha256-zk3Uk1SQyeRrQ7BuFwlOnQAptUHIkq+oPdfd+sTEq5U=",
+    "aarch64-darwin": "sha256-3BOd3EcqimoG3rTI6lTHe91YVlCoEi8/68eT1lbOi0c=",
+    "x86_64-darwin": "sha256-X7wGmjiMloF5Zhuc20kAxLC+tl613YNXRgA+dQjP2WM="
   }
 }

package.json

@@ -15,6 +15,7 @@
     "dev:stats": "bun sst shell --stage=production -- bun run --cwd packages/stats/app dev",
     "dev:www": "bun run --cwd packages/www dev",
     "dev:storybook": "bun --cwd packages/storybook storybook",
+    "bench:devex": "bun run --cwd packages/app test:bench:devex",
     "lint": "oxlint",
     "lint:effect-patterns": "ast-grep scan -c script/ast-grep/sgconfig.yml packages/util/src packages/core/src packages/server/src packages/protocol/src packages/cli/src",
     "test:lint-rules": "ast-grep test -c script/ast-grep/sgconfig.yml",
@@ -37,10 +38,10 @@
       "packages/slack"
     ],
     "catalog": {
-      "@effect/opentelemetry": "4.0.0-beta.107",
-      "@effect/platform-node": "4.0.0-beta.107",
-      "@effect/platform-node-shared": "4.0.0-beta.107",
-      "@effect/sql-sqlite-bun": "4.0.0-beta.107",
+      "@effect/opentelemetry": "4.0.0-rc.110",
+      "@effect/platform-node": "4.0.0-rc.110",
+      "@effect/platform-node-shared": "4.0.0-rc.110",
+      "@effect/sql-sqlite-bun": "4.0.0-rc.110",
       "@npmcli/arborist": "9.4.0",
       "@types/bun": "1.3.13",
       "@types/cross-spawn": "6.0.6",
@@ -71,7 +72,7 @@
       "dompurify": "3.3.1",
       "drizzle-kit": "1.0.0-rc.2",
       "drizzle-orm": "1.0.0-rc.2",
-      "effect": "4.0.0-beta.107",
+      "effect": "4.0.0-rc.110",
       "ai": "6.0.168",
       "cross-spawn": "7.0.6",
       "hono": "4.10.7",

packages/ai/README.md

@@ -309,7 +309,7 @@ Included providers: OpenAI, Anthropic, Google (Gemini), Google Vertex Gemini and
 
 ### Package-like entrypoints
 
-Native catalog integrations load provider behavior through package-like entrypoints. These are export paths from the same `@opencode-ai/ai` npm package, not independently published packages. Each entrypoint exports the same `model(modelID, settings)` contract, and `settings` contains serializable provider configuration plus common `headers`, `body`, and `limits` overlays.
+Native catalog integrations load provider behavior through package-like entrypoints. These are export paths from the same `@opencode-ai/ai` npm package, not independently published packages. Each entrypoint exports the same `model(modelID, settings)` contract, and `settings` contains serializable provider configuration plus common `headers` and `body` overlays.
 
 ```ts
 import { model } from "@opencode-ai/ai/providers/openai/responses"
@@ -317,7 +317,6 @@ import { model } from "@opencode-ai/ai/providers/openai/responses"
 const selected = model("gpt-5", {
   apiKey: process.env.OPENAI_API_KEY,
   headers: { "x-application": "opencode" },
-  limits: { context: 200_000, output: 64_000 },
 })
 ```
 

packages/ai/src/protocols/anthropic-messages.ts

@@ -31,6 +31,7 @@ import { ToolStream } from "./utils/tool-stream.js"
 const ADAPTER = "anthropic-messages"
 export const DEFAULT_BASE_URL = "https://api.anthropic.com/v1"
 export const PATH = "/messages"
+export const DEFAULT_MAX_TOKENS = 32_000
 
 export type ThinkingInput =
   | {
@@ -624,7 +625,6 @@ const resolveThinking = Effect.fn("AnthropicMessages.resolveThinking")(function*
 const fromRequest = Effect.fn("AnthropicMessages.fromRequest")(function* (request: LLMRequest) {
   const generation = request.generation
   const toolSchemaCompatibility = request.model.compatibility?.toolSchema
-  const outputLimit = request.model.defaults?.limits?.output ?? request.model.route.defaults.limits?.output ?? 4096
   // Allocate the 4-breakpoint budget in invalidation order: tools → system →
   // messages. Tools live highest in the cache hierarchy, so when callers
   // over-mark we keep their tool hints and shed the message-tail ones first.
@@ -663,7 +663,7 @@ const fromRequest = Effect.fn("AnthropicMessages.fromRequest")(function* (reques
     tools,
     tool_choice: toolChoice,
     stream: true as const,
-    max_tokens: generation?.maxTokens ?? outputLimit,
+    max_tokens: generation?.maxTokens ?? DEFAULT_MAX_TOKENS,
     temperature: generation?.temperature,
     top_p: generation?.topP,
     top_k: generation?.topK,

packages/ai/src/protocols/gemini.ts

@@ -289,7 +289,9 @@ const lowerMessages = Effect.fn("Gemini.lowerMessages")(function* (request: LLMR
     if (message.role === "system") {
       const part = yield* ProviderShared.wrappedSystemUpdate("Gemini", message)
       const previous = contents.at(-1)
-      if (previous?.role === "user")
+      // Gemini rejects a continuation whose function-response turn carries extra
+      // parts, so an update after a tool result starts its own user turn.
+      if (previous?.role === "user" && !previous.parts.some((item) => "functionResponse" in item))
         contents[contents.length - 1] = { role: "user", parts: [...previous.parts, { text: part.text }] }
       else contents.push({ role: "user", parts: [{ text: part.text }] })
       continue

packages/ai/src/protocols/open-responses.ts

@@ -93,6 +93,8 @@ export const InputItem = Schema.Union([
   Schema.Struct({ role: Schema.tag("developer"), content: Schema.String }),
   Schema.Struct({ role: Schema.tag("user"), content: Schema.Array(OpenResponsesInputContent) }),
   Schema.Struct({
+    type: Schema.tag("message"),
+    id: Schema.optionalKey(Schema.String),
     role: Schema.tag("assistant"),
     content: Schema.Array(OpenResponsesOutputText),
     phase: Schema.optionalKey(MessagePhase),
@@ -101,6 +103,7 @@ export const InputItem = Schema.Union([
   OpenResponsesItemReference,
   Schema.Struct({
     type: Schema.tag("function_call"),
+    id: Schema.optionalKey(Schema.String),
     call_id: Schema.String,
     name: Schema.String,
     arguments: Schema.String,
@@ -115,6 +118,8 @@ type OpenResponsesInputItem = Schema.Schema.Type<typeof InputItem>
 type LoweredInputItem =
   | OpenResponsesInputItem
   | {
+      readonly type: "message"
+      readonly id?: string
       readonly role: "assistant"
       readonly content: ReadonlyArray<{ readonly type: "output_text"; readonly text: string }>
       readonly phase?: MessagePhase | null
@@ -128,8 +133,6 @@ type OpenResponsesReasoningInput = {
   summary: Array<{ type: "summary_text"; text: string }>
   encrypted_content?: string | null
 }
-type OpenResponsesReasoningReplay = Omit<OpenResponsesReasoningInput, "id">
-
 export const Tool = Schema.Struct({
   type: Schema.tag("function"),
   name: Schema.String,
@@ -159,6 +162,14 @@ export const coreFields = {
   tools: optionalArray(Tool),
   tool_choice: Schema.optional(ToolChoice),
   store: Schema.optional(Schema.Boolean),
+  metadata: Schema.optional(Schema.Record(Schema.String, Schema.String)),
+  safety_identifier: Schema.optional(Schema.String),
+  stream_options: Schema.optional(
+    Schema.Struct({
+      include_obfuscation: Schema.optional(Schema.Boolean),
+    }),
+  ),
+  top_logprobs: Schema.optional(Schema.Int.check(Schema.isBetween({ minimum: 0, maximum: 20 }))),
   truncation: Schema.optional(OpenResponsesOptions.TruncationSchema),
   service_tier: Schema.optional(OpenResponsesOptions.ServiceTierSchema),
   prompt_cache_key: Schema.optional(Schema.String),
@@ -179,6 +190,8 @@ export const coreFields = {
   parallel_tool_calls: Schema.optional(Schema.Boolean),
   temperature: Schema.optional(Schema.Number),
   top_p: Schema.optional(Schema.Number),
+  presence_penalty: Schema.optional(Schema.Number),
+  frequency_penalty: Schema.optional(Schema.Number),
 }
 
 const OpenResponsesBody = Schema.Struct({
@@ -288,6 +301,20 @@ export const Event = Schema.StructWithRest(
 )
 export type Event = Schema.Schema.Type<typeof Event>
 
+const RefusalEvent = Schema.Union([
+  Schema.Struct({
+    type: Schema.tag("response.refusal.delta"),
+    item_id: Schema.String,
+    delta: Schema.String,
+  }),
+  Schema.Struct({
+    type: Schema.tag("response.refusal.done"),
+    item_id: Schema.String,
+    refusal: Schema.String,
+  }),
+])
+const isRefusalEvent = Schema.is(RefusalEvent)
+
 export interface Extension {
   readonly id: string
   readonly name: string
@@ -353,34 +380,42 @@ export const lowerToolChoice = (protocolName: string, toolChoice: NonNullable<LL
     tool: (toolName) => ({ type: "function" as const, name: toolName }),
   })
 
-const lowerToolCall = (part: ToolCallPart): OpenResponsesInputItem => ({
-  type: "function_call",
-  call_id: part.id,
-  name: part.name,
-  arguments: ProviderShared.encodeJson(part.input),
-})
+const itemID = (providerMetadata: ProviderMetadata | undefined, providerMetadataKey: string) => {
+  const metadata = providerMetadata?.[providerMetadataKey]
+  return ProviderShared.isRecord(metadata) && typeof metadata.itemId === "string" && metadata.itemId.length > 0
+    ? metadata.itemId
+    : undefined
+}
+
+const lowerToolCall = (part: ToolCallPart, providerMetadataKey: string): OpenResponsesInputItem => {
+  const id = itemID(part.providerMetadata, providerMetadataKey)
+  return {
+    type: "function_call",
+    ...(id ? { id } : {}),
+    call_id: part.id,
+    name: part.name,
+    arguments: ProviderShared.encodeJson(part.input),
+  }
+}
 
 const lowerReasoning = (part: ReasoningPart, providerMetadataKey: string): OpenResponsesReasoningInput | undefined => {
   const metadata = part.providerMetadata?.[providerMetadataKey]
-  if (!ProviderShared.isRecord(metadata) || typeof metadata.itemId !== "string" || metadata.itemId.length === 0)
-    return undefined
+  const id = itemID(part.providerMetadata, providerMetadataKey)
+  if (!ProviderShared.isRecord(metadata) || !id) return undefined
   const encryptedContent =
     typeof metadata.reasoningEncryptedContent === "string" || metadata.reasoningEncryptedContent === null
       ? metadata.reasoningEncryptedContent
       : undefined
   return {
     type: "reasoning",
-    id: metadata.itemId,
+    id,
     summary: part.text.length > 0 ? [{ type: "summary_text", text: part.text }] : [],
     encrypted_content: encryptedContent,
   }
 }
 
 const hostedToolItemID = (part: ToolResultPart, providerMetadataKey: string) => {
-  const metadata = part.providerMetadata?.[providerMetadataKey]
-  return ProviderShared.isRecord(metadata) && typeof metadata.itemId === "string" && metadata.itemId.length > 0
-    ? metadata.itemId
-    : undefined
+  return itemID(part.providerMetadata, providerMetadataKey)
 }
 
 const lowerMedia = Effect.fn("OpenResponses.lowerMedia")(function* (
@@ -465,24 +500,26 @@ const lowerMessages = Effect.fn("OpenResponses.lowerMessages")(function* (reques
 
     if (message.role === "assistant") {
       const content: TextPart[] = []
-      const reasoningItems: Record<string, OpenResponsesReasoningReplay> = {}
+      const reasoningItems: Record<string, OpenResponsesReasoningInput> = {}
       const reasoningReferences = new Set<string>()
       const hostedToolReferences = new Set<string>()
       const flushText = () => {
         if (content.length === 0) return
-        const groups = content.reduce<Array<{ phase: MessagePhase | null | undefined; parts: TextPart[] }>>(
-          (groups, part) => {
-            const metadata = part.providerMetadata?.[providerMetadataKey]
-            const phase = ProviderShared.isRecord(metadata) ? messagePhase(metadata.phase, extension) : undefined
-            const group = groups.at(-1)
-            if (group && group.phase === phase) group.parts.push(part)
-            else groups.push({ phase, parts: [part] })
-            return groups
-          },
-          [],
-        )
+        const groups = content.reduce<
+          Array<{ id: string | undefined; phase: MessagePhase | null | undefined; parts: TextPart[] }>
+        >((groups, part) => {
+          const metadata = part.providerMetadata?.[providerMetadataKey]
+          const id = itemID(part.providerMetadata, providerMetadataKey)
+          const phase = ProviderShared.isRecord(metadata) ? messagePhase(metadata.phase, extension) : undefined
+          const group = groups.at(-1)
+          if (group && group.id === id && group.phase === phase) group.parts.push(part)
+          else groups.push({ id, phase, parts: [part] })
+          return groups
+        }, [])
         input.push(
           ...groups.map((group) => ({
+            type: "message" as const,
+            ...(group.id === undefined ? {} : { id: group.id }),
             role: "assistant" as const,
             content: group.parts.map((part) => ({ type: "output_text" as const, text: part.text })),
             ...(group.phase === undefined ? {} : { phase: group.phase }),
@@ -511,19 +548,14 @@ const lowerMessages = Effect.fn("OpenResponses.lowerMessages")(function* (reques
               existing.encrypted_content = reasoning.encrypted_content
             continue
           }
-          const replay = {
-            type: reasoning.type,
-            summary: reasoning.summary,
-            encrypted_content: reasoning.encrypted_content,
-          }
-          reasoningItems[reasoning.id] = replay
-          input.push(replay)
+          reasoningItems[reasoning.id] = reasoning
+          input.push(reasoning)
           continue
         }
         if (part.type === "tool-call") {
           flushText()
           if (part.providerExecuted === true) continue
-          input.push(lowerToolCall(part))
+          input.push(lowerToolCall(part, providerMetadataKey))
           continue
         }
         if (part.type === "tool-result" && part.providerExecuted === true) {
@@ -578,6 +610,12 @@ const lowerOptions = (request: LLMRequest) => {
   return {
     ...(options.instructions ? { instructions: options.instructions } : {}),
     ...(options.store !== undefined ? { store: options.store } : {}),
+    ...(options.metadata ? { metadata: options.metadata } : {}),
+    ...(options.safetyIdentifier ? { safety_identifier: options.safetyIdentifier } : {}),
+    ...(options.streamOptions?.includeObfuscation !== undefined
+      ? { stream_options: { include_obfuscation: options.streamOptions.includeObfuscation } }
+      : {}),
+    ...(options.topLogprobs !== undefined ? { top_logprobs: options.topLogprobs } : {}),
     ...(request.promptCacheKey ? { prompt_cache_key: request.promptCacheKey } : {}),
     ...(options.include ? { include: options.include } : {}),
     ...(options.reasoningEffort || options.reasoningSummary
@@ -627,6 +665,8 @@ export const fromRequestWithExtension = Effect.fn("OpenResponses.fromRequestWith
     max_output_tokens: generation?.maxTokens,
     temperature: generation?.temperature,
     top_p: generation?.topP,
+    presence_penalty: generation?.presencePenalty,
+    frequency_penalty: generation?.frequencyPenalty,
     ...lowerOptions(request),
   }
 })
@@ -695,7 +735,7 @@ const onOutputTextDelta = (state: ParserState, event: Event, id: string): StepRe
   if (!event.delta) return [state, NO_EVENTS]
   const events: LLMEvent[] = []
   const phase = state.messagePhases[id]
-  const metadata = phase === undefined ? undefined : providerMetadata(state, { phase })
+  const metadata = providerMetadata(state, { itemId: id, ...(phase === undefined ? {} : { phase }) })
   const lifecycle = Lifecycle.textStart(state.lifecycle, events, id, metadata)
   return [{ ...state, lifecycle: Lifecycle.textDelta(lifecycle, events, id, event.delta) }, events]
 }
@@ -924,7 +964,7 @@ const onOutputItemDone = Effect.fn("OpenResponses.onOutputItemDone")(function* (
           state.lifecycle,
           events,
           item.id,
-          phase === undefined ? undefined : providerMetadata(state, { phase }),
+          providerMetadata(state, { itemId: item.id, ...(phase === undefined ? {} : { phase }) }),
         ),
         messageItems,
         messagePhases,
@@ -937,7 +977,11 @@ const onOutputItemDone = Effect.fn("OpenResponses.onOutputItemDone")(function* (
     if (!item.id || !item.call_id || !item.name) return [state, NO_EVENTS] satisfies StepResult
     const tools = state.tools[item.id]
       ? state.tools
-      : ToolStream.start(state.tools, item.id, { id: item.call_id, name: item.name })
+      : ToolStream.start(state.tools, item.id, {
+          id: item.call_id,
+          name: item.name,
+          providerMetadata: providerMetadata(state, { itemId: item.id }),
+        })
     const result =
       item.arguments === undefined
         ? yield* ToolStream.finish(state.id, tools, item.id)
@@ -988,11 +1032,19 @@ const onOutputItemDone = Effect.fn("OpenResponses.onOutputItemDone")(function* (
   return [state, NO_EVENTS] satisfies StepResult
 })
 
-const onResponseFinish = (state: ParserState, event: Event): StepResult => {
-  const events: LLMEvent[] = []
+const onResponseFinish = Effect.fn("OpenResponses.onResponseFinish")(function* (state: ParserState, event: Event) {
+  // Some compatible providers omit output_item.done even after completing the response.
+  const pending =
+    event.type === "response.completed"
+      ? yield* ToolStream.finishAll(state.id, state.tools)
+      : { tools: state.tools, events: NO_EVENTS }
+  const events: LLMEvent[] = [...pending.events]
+  const hasFunctionCall =
+    pending.events.some((event) => LLMEvent.is.toolCall(event) || LLMEvent.is.toolInputError(event)) ||
+    state.hasFunctionCall
   const lifecycle = Lifecycle.finish(state.lifecycle, events, {
     reason: {
-      normalized: mapFinishReason(event, state.hasFunctionCall),
+      normalized: mapFinishReason(event, hasFunctionCall),
       raw: event.response?.incomplete_details?.reason,
     },
     usage: mapUsage(event.response?.usage, state.providerMetadataKey),
@@ -1004,8 +1056,8 @@ const onResponseFinish = (state: ParserState, event: Event): StepResult => {
           })
         : undefined,
   })
-  return [{ ...state, lifecycle }, events]
-}
+  return [{ ...state, lifecycle, hasFunctionCall, tools: pending.tools }, events] satisfies StepResult
+})
 
 // Build a single human-readable message from whatever the provider supplied.
 // When both code and message are present, prefix the code so consumers see
@@ -1047,6 +1099,14 @@ export const step = (state: ParserState, event: Event) => {
         : onOutputTextDone(state, event, event.item_id),
     )
   }
+  if (event.type === "response.refusal.delta" || event.type === "response.refusal.done") {
+    if (!isRefusalEvent(event)) return ProviderShared.eventError(state.id, `${event.type} is malformed`)
+    return Effect.succeed(
+      event.type === "response.refusal.delta"
+        ? onOutputTextDelta(state, event, event.item_id)
+        : onOutputTextDone(state, { ...event, text: event.refusal }, event.item_id),
+    )
+  }
   if (event.type === "response.reasoning.delta" || event.type === "response.reasoning_summary_text.delta") {
     if (!event.item_id) return ProviderShared.eventError(state.id, `${event.type} is missing item_id`)
     return Effect.succeed(onReasoningDelta(state, event, event.item_id))
@@ -1074,8 +1134,7 @@ export const step = (state: ParserState, event: Event) => {
       return ProviderShared.eventError(state.id, `${event.type} message is missing id`)
     return onOutputItemDone(state, event)
   }
-  if (event.type === "response.completed" || event.type === "response.incomplete")
-    return Effect.succeed(onResponseFinish(state, event))
+  if (event.type === "response.completed" || event.type === "response.incomplete") return onResponseFinish(state, event)
   if (event.type === "response.failed") return providerError(state, event, `${state.name} response failed`)
   if (event.type === "error")
     return decodeKnownErrorEvent(event).pipe(

packages/ai/src/protocols/openai-chat.ts

@@ -28,7 +28,7 @@ import { ToolSchemaProjection } from "./utils/tool-schema.js"
 import { ToolStream } from "./utils/tool-stream.js"
 
 const ADAPTER = "openai-chat"
-const RESERVED_REASONING_FIELDS = new Set(["role", "content", "tool_calls"])
+const RESERVED_REASONING_FIELDS = new Set(["role", "content", "refusal", "tool_calls"])
 export const DEFAULT_BASE_URL = "https://api.openai.com/v1"
 export const PATH = "/chat/completions"
 
@@ -194,6 +194,7 @@ type OpenAIChatToolCallDelta = Schema.Schema.Type<typeof OpenAIChatToolCallDelta
 const OpenAIChatDelta = Schema.StructWithRest(
   Schema.Struct({
     content: optionalNull(Schema.String),
+    refusal: optionalNull(Schema.String),
     reasoning_content: optionalNull(Schema.String),
     reasoning: optionalNull(Schema.String),
     reasoning_text: optionalNull(Schema.String),
@@ -709,6 +710,7 @@ const step = (state: ParserState, event: OpenAIChatEvent) =>
     const reasoning = reasoningDelta(delta, state.reasoningField)
     const hasLateContent =
       Boolean(delta?.content) ||
+      Boolean(delta?.refusal) ||
       reasoning !== undefined ||
       (Array.isArray(delta?.reasoning_details) && delta.reasoning_details.length > 0) ||
       toolDeltas.some((tool) => Boolean(tool.id) || Boolean(tool.function?.name) || Boolean(tool.function?.arguments))
@@ -728,7 +730,7 @@ const step = (state: ParserState, event: OpenAIChatEvent) =>
     else if (
       reasoningDetailsObserved &&
       !lifecycle.reasoning.has("reasoning-0") &&
-      (Boolean(delta?.content) || toolDeltas.length > 0)
+      (Boolean(delta?.content) || Boolean(delta?.refusal) || toolDeltas.length > 0)
     )
       lifecycle = Lifecycle.reasoningStart(lifecycle, events, "reasoning-0", deltaMetadata)
     const reasoningEmitted = state.reasoningEmitted || lifecycle.reasoning.has("reasoning-0")
@@ -743,6 +745,16 @@ const step = (state: ParserState, event: OpenAIChatEvent) =>
       lifecycle = Lifecycle.textDelta(lifecycle, events, "text-0", delta.content)
     }
 
+    if (delta?.refusal) {
+      lifecycle = Lifecycle.reasoningEnd(
+        lifecycle,
+        events,
+        "reasoning-0",
+        reasoningMetadata(reasoningField, reasoningDetailsObserved ? state.reasoningDetails : undefined),
+      )
+      lifecycle = Lifecycle.textDelta(lifecycle, events, "text-0", delta.refusal)
+    }
+
     // Compatible providers may omit indexes. Prefer durable identity, then use
     // batch position for parallel deltas or the latest call for sparse chunks.
     for (const [position, tool] of toolDeltas.entries()) {

packages/ai/src/protocols/openai-responses.ts

@@ -42,6 +42,8 @@ const OpenAIResponsesToolChoice = Schema.Union([
 
 const OpenAIResponsesInputItem = Schema.Union([
   Schema.Struct({
+    type: Schema.tag("message"),
+    id: Schema.optionalKey(Schema.String),
     role: Schema.tag("assistant"),
     content: Schema.Array(Schema.Struct({ type: Schema.tag("output_text"), text: Schema.String })),
     phase: Schema.optionalKey(Schema.NullOr(OpenResponses.MessagePhase)),

packages/ai/src/protocols/utils/open-responses-options.ts

@@ -47,9 +47,17 @@ export const AllowedTools = Schema.Struct({
 })
 export type AllowedTools = typeof AllowedTools.Type
 
+export const StreamOptions = Schema.Struct({
+  includeObfuscation: Schema.optional(Schema.Boolean),
+})
+
 export const Options = Schema.Struct({
   instructions: Schema.optional(Schema.String),
   store: Schema.optional(Schema.Boolean),
+  metadata: Schema.optional(Schema.Record(Schema.String, Schema.String)),
+  safetyIdentifier: Schema.optional(Schema.String),
+  streamOptions: Schema.optional(StreamOptions),
+  topLogprobs: Schema.optional(Schema.Int.check(Schema.isBetween({ minimum: 0, maximum: 20 }))),
   reasoningEffort: Schema.optional(ReasoningEffort),
   reasoningSummary: Schema.optional(Schema.Literals(["auto", "concise", "detailed"])),
   include: Schema.optional(Schema.Array(ResponseIncludableSchema)),

packages/ai/src/provider-package.ts

@@ -4,11 +4,6 @@ export interface Settings extends Readonly<Record<string, unknown>> {
   readonly baseURL?: string
   readonly headers?: Readonly<Record<string, string>>
   readonly body?: Readonly<Record<string, unknown>>
-  readonly limits?: {
-    readonly context: number
-    readonly input?: number
-    readonly output: number
-  }
 }
 
 export interface Definition<

packages/ai/src/providers/amazon-bedrock-mantle.ts

@@ -90,7 +90,6 @@ const config = (settings: Settings): Config => {
     credentials: settings.credentials,
     headers: settings.headers === undefined ? undefined : { ...settings.headers },
     http: settings.body === undefined ? undefined : { body: { ...settings.body } },
-    limits: settings.limits,
     providerOptions: settings.providerOptions,
     region: settings.region,
   }

packages/ai/src/providers/amazon-bedrock.ts

@@ -62,7 +62,6 @@ export const model: ProviderPackage.Definition<Settings>["model"] = (modelID, se
     generation: settings.topP === undefined ? undefined : { topP: settings.topP },
     headers: settings.headers === undefined ? undefined : { ...settings.headers },
     http: settings.body === undefined ? undefined : { body: { ...settings.body } },
-    limits: settings.limits,
     region: settings.region,
   }).model(modelID)
 }

packages/ai/src/providers/anthropic-compatible.ts

@@ -68,7 +68,6 @@ export const model: ProviderPackage.Definition<Settings, AnthropicMessages.Provi
     baseURL: settings.baseURL,
     headers: settings.headers === undefined ? undefined : { ...settings.headers },
     http: settings.body === undefined ? undefined : { body: { ...settings.body } },
-    limits: settings.limits,
     provider: settings.provider,
     providerOptions: settings.providerOptions,
   }).model(modelID)

packages/ai/src/providers/anthropic.ts

@@ -63,7 +63,6 @@ export const model: ProviderPackage.Definition<Settings, AnthropicMessages.Provi
     baseURL: settings.baseURL,
     headers: settings.headers === undefined ? undefined : { ...settings.headers },
     http: settings.body === undefined ? undefined : { body: { ...settings.body } },
-    limits: settings.limits,
     providerOptions: settings.providerOptions,
   }).model(modelID)
 }

packages/ai/src/providers/azure.ts

@@ -126,7 +126,6 @@ const config = (settings: Settings): Config => {
     apiVersion: settings.apiVersion,
     headers: settings.headers === undefined ? undefined : { ...settings.headers },
     http: settings.body === undefined ? undefined : { body: { ...settings.body } },
-    limits: settings.limits,
     providerOptions: settings.providerOptions,
     queryParams: settings.queryParams === undefined ? undefined : { ...settings.queryParams },
     useDeploymentBasedUrls: settings.useDeploymentBasedUrls,

packages/ai/src/providers/google-vertex-chat.ts

@@ -75,7 +75,6 @@ export const model: ProviderPackage.Definition<Settings, OpenAIProviderOptionsIn
     baseURL: settings.baseURL,
     headers: settings.headers === undefined ? undefined : { ...settings.headers },
     http: settings.body === undefined ? undefined : { body: { ...settings.body } },
-    limits: settings.limits,
     location: settings.location,
     project: settings.project,
     providerOptions: settings.providerOptions,

packages/ai/src/providers/google-vertex-messages.ts

@@ -110,7 +110,6 @@ export const model: ProviderPackage.Definition<Settings, AnthropicMessages.Provi
     baseURL: settings.baseURL,
     headers: settings.headers === undefined ? undefined : { ...settings.headers },
     http: settings.body === undefined ? undefined : { body: { ...settings.body } },
-    limits: settings.limits,
     location: settings.location,
     project: settings.project,
     providerOptions: settings.providerOptions,

packages/ai/src/providers/google-vertex-responses.ts

@@ -80,7 +80,6 @@ export const model: ProviderPackage.Definition<Settings, OpenResponsesProviderOp
     baseURL: settings.baseURL,
     headers: settings.headers === undefined ? undefined : { ...settings.headers },
     http: settings.body === undefined ? undefined : { body: { ...settings.body } },
-    limits: settings.limits,
     location: settings.location,
     project: settings.project,
     providerOptions: settings.providerOptions,

packages/ai/src/providers/google-vertex.ts

@@ -119,7 +119,6 @@ export const model: ProviderPackage.Definition<Settings, GeminiProviderOptionsIn
     baseURL: settings.baseURL,
     headers: settings.headers === undefined ? undefined : { ...settings.headers },
     http: settings.body === undefined ? undefined : { body: { ...settings.body } },
-    limits: settings.limits,
     location: settings.location,
     project: settings.project,
     providerOptions: settings.providerOptions,

packages/ai/src/providers/google.ts

@@ -63,7 +63,6 @@ export const model: ProviderPackage.Definition<Settings, Gemini.ProviderOptionsI
     baseURL: settings.baseURL,
     headers: settings.headers === undefined ? undefined : { ...settings.headers },
     http: settings.body === undefined ? undefined : { body: { ...settings.body } },
-    limits: settings.limits,
     providerOptions: settings.providerOptions,
   }).model(modelID)
 

packages/ai/src/providers/openai-compatible-responses.ts

@@ -55,7 +55,6 @@ export const model: ProviderPackage.Definition<Settings, OpenResponsesProviderOp
     baseURL: settings.baseURL,
     headers: settings.headers === undefined ? undefined : { ...settings.headers },
     http: settings.body === undefined ? undefined : { body: { ...settings.body } },
-    limits: settings.limits,
     provider: settings.provider,
     providerOptions: settings.providerOptions,
   }).model(modelID)

packages/ai/src/providers/openai-compatible.ts

@@ -74,7 +74,6 @@ export const model: ProviderPackage.Definition<Settings, OpenAIProviderOptionsIn
     baseURL: settings.baseURL,
     headers: settings.headers === undefined ? undefined : { ...settings.headers },
     http: settings.body === undefined ? undefined : { body: { ...settings.body } },
-    limits: settings.limits,
     provider: settings.provider,
     providerOptions: settings.providerOptions,
   }).model(modelID)

packages/ai/src/providers/openai.ts

@@ -124,7 +124,6 @@ const config = (settings: Settings): Config => {
     baseURL: settings.baseURL,
     headers: Object.keys(headers).length === 0 ? undefined : headers,
     http: settings.body === undefined ? undefined : { body: { ...settings.body } },
-    limits: settings.limits,
     providerOptions: settings.providerOptions,
     queryParams: settings.queryParams === undefined ? undefined : { ...settings.queryParams },
   }

packages/ai/src/providers/openrouter.ts

@@ -200,6 +200,5 @@ export const model: ProviderPackage.Definition<Settings, OpenRouterProviderOptio
     baseURL: settings.baseURL,
     headers: settings.headers,
     http: settings.body === undefined ? undefined : { body: { ...settings.body } },
-    limits: settings.limits,
     providerOptions: settings.providerOptions,
   }).model(modelID)

packages/ai/src/providers/xai.ts

@@ -101,7 +101,6 @@ export const model: ProviderPackage.Definition<Settings, XAIProviderOptionsInput
     baseURL: settings.baseURL,
     headers: settings.headers,
     http: settings.body === undefined ? undefined : { body: { ...settings.body } },
-    limits: settings.limits,
     providerOptions: settings.providerOptions,
   }).model(modelID)
 export const responses = provider.responses

packages/ai/src/route/client.ts

@@ -16,7 +16,6 @@ import {
   LLMRequest,
   LLMResponse,
   LanguageModel,
-  LanguageModelLimits,
   LLMEvent,
   InvalidProviderOutputReason,
   ProviderID,
@@ -74,7 +73,6 @@ export type RouteRoutedLanguageModelInput = Omit<LanguageModel.Input, "route">
 
 export interface RouteDefaults {
   readonly headers?: Record<string, string>
-  readonly limits?: LanguageModelLimits
   readonly generation?: GenerationOptions
   readonly providerOptions?: ProviderOptions
   readonly http?: HttpOptions
@@ -82,7 +80,6 @@ export interface RouteDefaults {
 
 export interface RouteDefaultsInput {
   readonly headers?: Record<string, string>
-  readonly limits?: LanguageModelLimits.Input
   readonly generation?: GenerationOptions.Input
   readonly providerOptions?: ProviderOptions
   readonly http?: HttpOptions.Input
@@ -119,7 +116,6 @@ const mergeRouteDefaults = (base: RouteDefaults | undefined, patch: RouteDefault
     ...base,
     ...patch,
     headers,
-    limits: patch.limits === undefined ? base?.limits : LanguageModelLimits.make(patch.limits),
     generation: mergeGenerationOptions(generationOptions(base?.generation), generationOptions(patch.generation)),
     providerOptions: mergeProviderOptions(base?.providerOptions, patch.providerOptions),
     http: mergeHttpOptions(

packages/ai/src/route/executor.ts

@@ -188,12 +188,14 @@ const statusError =
 
 // Classifies an HTTP failure captured outside the executor (for example by the
 // AI SDK's own fetch) onto the same reason types and HttpContext that
-// executor-driven requests produce. The originating request is not available on
-// that path, so the method is assumed (language model calls are always POST),
-// request headers are empty.
+// executor-driven requests produce. Callers may supply originating request
+// details when their transport exposes them; language model calls otherwise
+// default to POST with no request headers.
 export const classifyHttpFailure = (input: {
   readonly message: string
   readonly url: string
+  readonly method?: string | undefined
+  readonly requestHeaders?: Record<string, string> | undefined
   readonly status?: number | undefined
   readonly code?: string | undefined
   readonly responseHeaders?: Record<string, string> | undefined
@@ -210,7 +212,11 @@ export const classifyHttpFailure = (input: {
     retryAfterMs: retryAfter,
     rateLimit,
     http: new HttpContext({
-      request: new HttpRequestDetails({ method: "POST", url: input.url, headers: {} }),
+      request: new HttpRequestDetails({
+        method: input.method ?? "POST",
+        url: input.url,
+        headers: input.requestHeaders ?? {},
+      }),
       response:
         input.status === undefined
           ? undefined

packages/ai/src/route/transport/websocket.ts

@@ -2,6 +2,7 @@ import { Cause, Effect, Queue, Stream } from "effect"
 import { Headers } from "effect/unstable/http"
 import { Socket } from "effect/unstable/socket"
 import { AIError, TransportReason, type TransportOperation } from "../../schema/index.js"
+import { classifyHttpFailure } from "../executor.js"
 import * as HttpTransport from "./http.js"
 import type { Transport } from "./index.js"
 import type {
@@ -31,6 +32,18 @@ type WebSocketConstructorWithHeaders = (
   options?: { readonly headers?: Headers.Headers },
 ) => globalThis.WebSocket
 
+interface UpgradeResponse extends AsyncIterable<unknown> {
+  readonly statusCode?: number
+  readonly headers: Readonly<Record<string, string | ReadonlyArray<string> | undefined>>
+}
+
+type UpgradeListener = (request: unknown, response: UpgradeResponse) => void
+
+interface UpgradeAwareWebSocket extends globalThis.WebSocket {
+  readonly once?: (event: "unexpected-response", listener: UpgradeListener) => void
+  readonly off?: (event: "unexpected-response", listener: UpgradeListener) => void
+}
+
 const MAX_FRAME_BYTES = 16 * 1024 * 1024
 const transportError = (
   method: string,
@@ -91,6 +104,59 @@ const binaryMessage = (data: unknown) => {
   return undefined
 }
 
+const upgradeHeaders = (headers: UpgradeResponse["headers"]) =>
+  Object.fromEntries(
+    Object.entries(headers).flatMap(([name, value]) => {
+      if (value === undefined) return []
+      return [[name, Array.isArray(value) ? value.join(", ") : String(value)]]
+    }),
+  )
+
+const upgradeBody = (input: WebSocketRequest, response: UpgradeResponse) =>
+  Stream.fromAsyncIterable(response, (error) =>
+    transportError("open", error instanceof Error ? error.message : "Failed to read WebSocket upgrade response", {
+      url: input.url,
+      operation: "read",
+      phase: "connect",
+      delivery: "not-sent",
+    }),
+  ).pipe(
+    Stream.mapEffect((chunk) => {
+      if (typeof chunk === "string") return Effect.succeed(new TextEncoder().encode(chunk))
+      const binary = binaryMessage(chunk)
+      if (binary) return Effect.succeed(binary)
+      return Effect.fail(
+        transportError("open", "Unsupported WebSocket upgrade response body", {
+          url: input.url,
+          operation: "read",
+          phase: "connect",
+          delivery: "not-sent",
+        }),
+      )
+    }),
+    Stream.decodeText(),
+    Stream.runFold(
+      () => "",
+      (body, chunk) => body + chunk,
+    ),
+  )
+
+const rejectedUpgrade = (input: WebSocketRequest, response: UpgradeResponse, body: string | undefined) =>
+  new AIError({
+    module: "WebSocketConnector",
+    method: "open",
+    reason: classifyHttpFailure({
+      message: `WebSocket upgrade rejected with HTTP ${response.statusCode ?? "unknown"}`,
+      method: "GET",
+      url: input.url,
+      requestHeaders: { ...input.headers },
+      status: response.statusCode,
+      code: "UnexpectedServerResponse",
+      responseHeaders: upgradeHeaders(response.headers),
+      responseBody: body,
+    }),
+  })
+
 const waitOpen = (ws: globalThis.WebSocket, input: WebSocketRequest) => {
   if (ws.readyState === globalThis.WebSocket.OPEN) return Effect.void
   if (ws.readyState === globalThis.WebSocket.CLOSING || ws.readyState === globalThis.WebSocket.CLOSED) {
@@ -105,10 +171,13 @@ const waitOpen = (ws: globalThis.WebSocket, input: WebSocketRequest) => {
     )
   }
   return Effect.callback<void, AIError>((resume, signal) => {
+    const upgrade: UpgradeAwareWebSocket = ws
+    const unexpectedResponse = upgrade.once && upgrade.off ? upgrade : undefined
     const cleanup = () => {
       ws.removeEventListener("open", onOpen)
       ws.removeEventListener("error", onError)
       ws.removeEventListener("close", onClose)
+      unexpectedResponse?.off?.("unexpected-response", onUnexpectedResponse)
       signal.removeEventListener("abort", onAbort)
     }
     const onAbort = () => {
@@ -147,9 +216,25 @@ const waitOpen = (ws: globalThis.WebSocket, input: WebSocketRequest) => {
         ),
       )
     }
+    const onUnexpectedResponse: UpgradeListener = (_request, response) => {
+      cleanup()
+      resume(
+        upgradeBody(input, response).pipe(
+          Effect.catch(() => Effect.succeed(undefined)),
+          Effect.flatMap((body) => Effect.fail(rejectedUpgrade(input, response, body))),
+          Effect.ensuring(
+            Effect.sync(() => {
+              ws.addEventListener("error", () => {}, { once: true })
+              if (ws.readyState === globalThis.WebSocket.CONNECTING) ws.close()
+            }),
+          ),
+        ),
+      )
+    }
     ws.addEventListener("open", onOpen, { once: true })
     ws.addEventListener("error", onError, { once: true })
     ws.addEventListener("close", onClose, { once: true })
+    unexpectedResponse?.once?.("unexpected-response", onUnexpectedResponse)
     signal.addEventListener("abort", onAbort, { once: true })
   })
 }

packages/ai/src/schema/options.ts

@@ -114,22 +114,7 @@ export const mergeGenerationOptions = (...items: ReadonlyArray<GenerationOptions
   return Object.values(result).some((value) => value !== undefined) ? result : undefined
 }
 
-export class LanguageModelLimits extends Schema.Class<LanguageModelLimits>("LLM.LanguageModelLimits")({
-  context: Schema.optional(Schema.Number),
-  input: Schema.optional(Schema.Number),
-  output: Schema.optional(Schema.Number),
-}) {}
-
-export namespace LanguageModelLimits {
-  export type Input = LanguageModelLimits | ConstructorParameters<typeof LanguageModelLimits>[0]
-
-  /** Normalize model limit input into the canonical `LanguageModelLimits` class. */
-  export const make = (input: Input | undefined) =>
-    input instanceof LanguageModelLimits ? input : new LanguageModelLimits(input ?? {})
-}
-
 export class LanguageModelDefaults extends Schema.Class<LanguageModelDefaults>("LLM.LanguageModelDefaults")({
-  limits: Schema.optional(LanguageModelLimits),
   generation: Schema.optional(GenerationOptions),
   providerOptions: Schema.optional(ProviderOptions),
   http: Schema.optional(HttpOptions),
@@ -139,7 +124,6 @@ export namespace LanguageModelDefaults {
   export type Input =
     | LanguageModelDefaults
     | {
-        readonly limits?: LanguageModelLimits.Input
         readonly generation?: GenerationOptions.Input
         readonly providerOptions?: ProviderOptions
         readonly http?: HttpOptions.Input
@@ -149,7 +133,6 @@ export namespace LanguageModelDefaults {
   export const make = (input: Input) => {
     if (input instanceof LanguageModelDefaults) return input
     return new LanguageModelDefaults({
-      limits: input.limits === undefined ? undefined : LanguageModelLimits.make(input.limits),
       generation: input.generation === undefined ? undefined : GenerationOptions.make(input.generation),
       providerOptions: input.providerOptions,
       http: input.http === undefined ? undefined : HttpOptions.make(input.http),

packages/ai/test/compile.test.ts

@@ -270,21 +270,20 @@ describe("request option precedence", () => {
     ),
   )
 
-  it.effect("uses model output limits after route limits and before call maxTokens", () =>
+  it.effect("uses the Anthropic default before call maxTokens", () =>
     Effect.gen(function* () {
       const route = AnthropicMessages.route.with({
         endpoint: { baseURL: "https://api.anthropic.test/v1/" },
         auth: Auth.header("x-api-key", "test"),
-        limits: { output: 128 },
       })
-      const model = route.model({ id: "claude-sonnet-4-5", defaults: { limits: { output: 64 } } })
+      const model = route.model({ id: "claude-sonnet-4-5" })
       const withoutMaxTokens = yield* compileRequest(LLM.request({ model, prompt: "Say hello.", cache: "none" }))
       const withMaxTokens = yield* compileRequest(
-        LLM.request({ model, prompt: "Say hello.", cache: "none", generation: { maxTokens: 32 } }),
+        LLM.request({ model, prompt: "Say hello.", cache: "none", generation: { maxTokens: 8_000 } }),
       )
 
-      expect(withoutMaxTokens.body.max_tokens).toBe(64)
-      expect(withMaxTokens.body.max_tokens).toBe(32)
+      expect(withoutMaxTokens.body.max_tokens).toBe(32_000)
+      expect(withMaxTokens.body.max_tokens).toBe(8_000)
     }),
   )
 })

packages/ai/test/executor.test.ts

@@ -1,6 +1,7 @@
 import { describe, expect } from "bun:test"
 import { Deferred, Effect, Fiber, Layer, Ref, Stream } from "effect"
 import { Headers, HttpClient, HttpClientError, HttpClientRequest, HttpClientResponse } from "effect/unstable/http"
+import { Socket } from "effect/unstable/socket"
 import { LLM, AIError } from "../src/index.js"
 import { LLMClient, RequestExecutor, WebSocketTransport, type WebSocketChannelExecutor } from "../src/route.js"
 import * as OpenAIChat from "../src/protocols/openai-chat.js"
@@ -66,6 +67,53 @@ const expectAIError = (error: unknown) => {
 }
 
 const errorHttp = (error: AIError) => ("http" in error.reason ? error.reason.http : undefined)
+
+interface RejectedUpgradeResponse extends AsyncIterable<Uint8Array> {
+  readonly statusCode: number
+  readonly headers: Readonly<Record<string, string | undefined>>
+}
+
+type RejectedUpgradeListener = (request: unknown, response: RejectedUpgradeResponse) => void
+
+class RejectedUpgradeSocket extends EventTarget {
+  readyState = globalThis.WebSocket.CONNECTING
+  listener?: RejectedUpgradeListener
+
+  constructor(
+    readonly response: { status: number; headers: Readonly<Record<string, string | undefined>>; body: string },
+  ) {
+    super()
+  }
+
+  once(_event: "unexpected-response", listener: RejectedUpgradeListener) {
+    this.listener = listener
+    const body = this.response.body
+    queueMicrotask(() =>
+      listener(
+        {},
+        {
+          statusCode: this.response.status,
+          headers: this.response.headers,
+          async *[Symbol.asyncIterator]() {
+            const split = Math.floor(body.length / 2)
+            yield new TextEncoder().encode(body.slice(0, split))
+            yield new TextEncoder().encode(body.slice(split))
+          },
+        },
+      ),
+    )
+  }
+
+  off(_event: "unexpected-response", listener: RejectedUpgradeListener) {
+    if (this.listener === listener) this.listener = undefined
+  }
+
+  close() {
+    this.readyState = globalThis.WebSocket.CLOSED
+  }
+
+  send() {}
+}
 const largeProviderMessage = `Upstream request failed: ${"validation failed; ".repeat(1_000)}`
 
 describe("RequestExecutor", () => {
@@ -622,6 +670,78 @@ describe("WebSocket channel execution", () => {
     }),
   )
 
+  it.effect("preserves rejected upgrade diagnostics", () =>
+    Effect.gen(function* () {
+      const cases = [
+        {
+          status: 401,
+          headers: { "x-request-id": "req_401" },
+          body: '{"error":{"message":"invalid key"}}',
+          reason: { _tag: "Authentication", kind: "invalid" },
+        },
+        {
+          status: 403,
+          headers: { "x-request-id": "req_403" },
+          body: '{"error":{"message":"forbidden"}}',
+          reason: { _tag: "Authentication", kind: "insufficient-permissions" },
+        },
+        {
+          status: 429,
+          headers: {
+            "retry-after": "2",
+            "x-request-id": "req_429",
+            "x-ratelimit-limit-requests": "500",
+          },
+          body: '{"error":{"message":"rate limited"}}',
+          reason: { _tag: "RateLimit", retryAfterMs: 2_000 },
+        },
+        {
+          status: 503,
+          headers: { "retry-after-ms": "250", "x-request-id": "req_503" },
+          body: '{"error":{"message":"overloaded"}}',
+          reason: { _tag: "ProviderInternal", status: 503, retryAfterMs: 250 },
+        },
+      ] as const
+
+      yield* Effect.forEach(
+        cases,
+        (item) =>
+          Effect.gen(function* () {
+            const socket = new RejectedUpgradeSocket(item)
+            const error = yield* WebSocketTransport.open({
+              url: "wss://api.openai.test/v1/responses",
+              headers: Headers.fromInput({ authorization: "Bearer secret", "x-client": "visible" }),
+            }).pipe(
+              Effect.provideService(Socket.WebSocketConstructor, () => {
+                // oxlint-disable-next-line typescript-eslint/no-unsafe-type-assertion -- fixture implements the socket surface used by the connector.
+                return socket as unknown as globalThis.WebSocket
+              }),
+              Effect.flip,
+            )
+
+            expectAIError(error)
+            expect(error.reason).toMatchObject(item.reason)
+            expect(errorHttp(error)).toMatchObject({
+              request: {
+                method: "GET",
+                url: "wss://api.openai.test/v1/responses",
+                headers: { authorization: "Bearer secret", "x-client": "visible" },
+              },
+              response: { status: item.status, headers: item.headers },
+              body: item.body,
+              requestId: `req_${item.status}`,
+              ...(item.status === 429
+                ? { rateLimit: { retryAfterMs: 2_000, limit: { requests: "500" } } }
+                : item.status === 503
+                  ? { rateLimit: { retryAfterMs: 250 } }
+                  : {}),
+            })
+          }),
+        { discard: true },
+      )
+    }),
+  )
+
   it.effect("uses HTTP when no per-call WebSocket executor is provided", () =>
     Effect.gen(function* () {
       const response = yield* LLMClient.generate(request).pipe(Effect.provide(fixedResponse(sseEvents(...frames))))

packages/ai/test/fixtures/recordings/openai-responses-websocket/continues-a-tool-call-over-one-socket.json

@@ -0,0 +1,100 @@
+{
+  "version": 1,
+  "metadata": {
+    "provider": "openai",
+    "protocol": "openai-responses",
+    "transport": "websocket",
+    "model": "gpt-5.5",
+    "tags": [
+      "prefix:openai-responses-websocket",
+      "provider:openai",
+      "protocol:openai-responses",
+      "transport:websocket",
+      "tool",
+      "continuation"
+    ],
+    "name": "openai-responses-websocket/continues-a-tool-call-over-one-socket",
+    "recordedAt": "2026-08-20T00:00:00.000Z"
+  },
+  "interactions": [
+    {
+      "transport": "websocket",
+      "connection": {
+        "sequence": 0,
+        "url": "wss://api.openai.com/v1/responses",
+        "protocols": [],
+        "close": {
+          "code": 1000,
+          "reason": ""
+        }
+      },
+      "events": [
+        {
+          "direction": "client",
+          "kind": "text",
+          "body": "{\"type\":\"response.create\",\"model\":\"gpt-5.5\",\"input\":[{\"role\":\"system\",\"content\":\"Call get_weather once, then reply exactly: Paris is sunny.\"},{\"role\":\"user\",\"content\":[{\"type\":\"input_text\",\"text\":\"What is the weather in Paris?\"}]}],\"tools\":[{\"type\":\"function\",\"name\":\"get_weather\",\"description\":\"Get current weather for a city.\",\"parameters\":{\"type\":\"object\",\"properties\":{\"city\":{\"type\":\"string\"}},\"required\":[\"city\"],\"additionalProperties\":false},\"strict\":false}],\"store\":false,\"max_output_tokens\":50,\"include\":[\"reasoning.encrypted_content\"],\"reasoning\":{\"effort\":\"medium\",\"summary\":\"auto\"},\"text\":{\"verbosity\":\"low\"}}"
+        },
+        {
+          "direction": "server",
+          "kind": "text",
+          "body": "{\"type\":\"response.created\",\"response\":{\"id\":\"resp_ws_tool_1\"}}"
+        },
+        {
+          "direction": "server",
+          "kind": "text",
+          "body": "{\"type\":\"response.output_item.added\",\"item\":{\"type\":\"function_call\",\"id\":\"fc_ws_weather\",\"call_id\":\"call_ws_weather\",\"name\":\"get_weather\",\"arguments\":\"\"}}"
+        },
+        {
+          "direction": "server",
+          "kind": "text",
+          "body": "{\"type\":\"response.function_call_arguments.delta\",\"item_id\":\"fc_ws_weather\",\"delta\":\"{\\\"city\\\":\\\"Paris\\\"}\"}"
+        },
+        {
+          "direction": "server",
+          "kind": "text",
+          "body": "{\"type\":\"response.output_item.done\",\"item\":{\"type\":\"function_call\",\"id\":\"fc_ws_weather\",\"call_id\":\"call_ws_weather\",\"name\":\"get_weather\",\"arguments\":\"{\\\"city\\\":\\\"Paris\\\"}\"}}"
+        },
+        {
+          "direction": "server",
+          "kind": "text",
+          "body": "{\"type\":\"response.completed\",\"response\":{\"id\":\"resp_ws_tool_1\"}}"
+        },
+        {
+          "direction": "client",
+          "kind": "text",
+          "body": "{\"type\":\"response.create\",\"model\":\"gpt-5.5\",\"input\":[{\"type\":\"function_call_output\",\"call_id\":\"call_ws_weather\",\"output\":\"{\\\"temperature\\\":22,\\\"condition\\\":\\\"sunny\\\"}\"}],\"tools\":[{\"type\":\"function\",\"name\":\"get_weather\",\"description\":\"Get current weather for a city.\",\"parameters\":{\"type\":\"object\",\"properties\":{\"city\":{\"type\":\"string\"}},\"required\":[\"city\"],\"additionalProperties\":false},\"strict\":false}],\"store\":false,\"max_output_tokens\":50,\"previous_response_id\":\"resp_ws_tool_1\",\"include\":[\"reasoning.encrypted_content\"],\"reasoning\":{\"effort\":\"medium\",\"summary\":\"auto\"},\"text\":{\"verbosity\":\"low\"}}"
+        },
+        {
+          "direction": "server",
+          "kind": "text",
+          "body": "{\"type\":\"response.created\",\"response\":{\"id\":\"resp_ws_tool_2\"}}"
+        },
+        {
+          "direction": "server",
+          "kind": "text",
+          "body": "{\"type\":\"response.output_item.added\",\"item\":{\"type\":\"message\",\"id\":\"msg_ws_tool_2\",\"role\":\"assistant\",\"content\":[]}}"
+        },
+        {
+          "direction": "server",
+          "kind": "text",
+          "body": "{\"type\":\"response.output_text.delta\",\"item_id\":\"msg_ws_tool_2\",\"delta\":\"Paris is sunny.\"}"
+        },
+        {
+          "direction": "server",
+          "kind": "text",
+          "body": "{\"type\":\"response.output_text.done\",\"item_id\":\"msg_ws_tool_2\",\"text\":\"Paris is sunny.\"}"
+        },
+        {
+          "direction": "server",
+          "kind": "text",
+          "body": "{\"type\":\"response.output_item.done\",\"item\":{\"type\":\"message\",\"id\":\"msg_ws_tool_2\",\"role\":\"assistant\",\"content\":[{\"type\":\"output_text\",\"text\":\"Paris is sunny.\"}]}}"
+        },
+        {
+          "direction": "server",
+          "kind": "text",
+          "body": "{\"type\":\"response.completed\",\"response\":{\"id\":\"resp_ws_tool_2\"}}"
+        }
+      ]
+    }
+  ]
+}

packages/ai/test/fixtures/recordings/openai-responses-websocket/reconstructs-full-context-after-reconnect.json

@@ -0,0 +1,119 @@
+{
+  "version": 1,
+  "metadata": {
+    "provider": "openai",
+    "protocol": "openai-responses",
+    "transport": "websocket",
+    "model": "gpt-5.5",
+    "tags": [
+      "prefix:openai-responses-websocket",
+      "provider:openai",
+      "protocol:openai-responses",
+      "transport:websocket",
+      "reconnect",
+      "full-context"
+    ],
+    "name": "openai-responses-websocket/reconstructs-full-context-after-reconnect",
+    "recordedAt": "2026-08-20T00:00:00.000Z"
+  },
+  "interactions": [
+    {
+      "transport": "websocket",
+      "connection": {
+        "sequence": 0,
+        "url": "wss://api.openai.com/v1/responses",
+        "protocols": [],
+        "close": {
+          "code": 1000,
+          "reason": ""
+        }
+      },
+      "events": [
+        {
+          "direction": "client",
+          "kind": "text",
+          "body": "{\"type\":\"response.create\",\"model\":\"gpt-5.5\",\"input\":[{\"role\":\"system\",\"content\":\"Follow the user's exact reply instruction.\"},{\"role\":\"user\",\"content\":[{\"type\":\"input_text\",\"text\":\"Reply exactly: Alpha.\"}]}],\"store\":false,\"max_output_tokens\":30,\"include\":[\"reasoning.encrypted_content\"],\"reasoning\":{\"effort\":\"medium\",\"summary\":\"auto\"},\"text\":{\"verbosity\":\"low\"}}"
+        },
+        {
+          "direction": "server",
+          "kind": "text",
+          "body": "{\"type\":\"response.created\",\"response\":{\"id\":\"resp_ws_reconnect_1\"}}"
+        },
+        {
+          "direction": "server",
+          "kind": "text",
+          "body": "{\"type\":\"response.output_item.added\",\"item\":{\"type\":\"message\",\"id\":\"msg_ws_reconnect_1\",\"role\":\"assistant\",\"content\":[]}}"
+        },
+        {
+          "direction": "server",
+          "kind": "text",
+          "body": "{\"type\":\"response.output_text.delta\",\"item_id\":\"msg_ws_reconnect_1\",\"delta\":\"Alpha.\"}"
+        },
+        {
+          "direction": "server",
+          "kind": "text",
+          "body": "{\"type\":\"response.output_text.done\",\"item_id\":\"msg_ws_reconnect_1\",\"text\":\"Alpha.\"}"
+        },
+        {
+          "direction": "server",
+          "kind": "text",
+          "body": "{\"type\":\"response.output_item.done\",\"item\":{\"type\":\"message\",\"id\":\"msg_ws_reconnect_1\",\"role\":\"assistant\",\"content\":[{\"type\":\"output_text\",\"text\":\"Alpha.\"}]}}"
+        },
+        {
+          "direction": "server",
+          "kind": "text",
+          "body": "{\"type\":\"response.completed\",\"response\":{\"id\":\"resp_ws_reconnect_1\"}}"
+        }
+      ]
+    },
+    {
+      "transport": "websocket",
+      "connection": {
+        "sequence": 1,
+        "url": "wss://api.openai.com/v1/responses",
+        "protocols": [],
+        "close": {
+          "code": 1000,
+          "reason": ""
+        }
+      },
+      "events": [
+        {
+          "direction": "client",
+          "kind": "text",
+          "body": "{\"type\":\"response.create\",\"model\":\"gpt-5.5\",\"input\":[{\"role\":\"system\",\"content\":\"Follow the user's exact reply instruction.\"},{\"role\":\"user\",\"content\":[{\"type\":\"input_text\",\"text\":\"Reply exactly: Alpha.\"}]},{\"type\":\"message\",\"id\":\"msg_ws_reconnect_1\",\"role\":\"assistant\",\"content\":[{\"type\":\"output_text\",\"text\":\"Alpha.\"}]},{\"role\":\"user\",\"content\":[{\"type\":\"input_text\",\"text\":\"Reply exactly: Beta.\"}]}],\"store\":false,\"max_output_tokens\":30,\"include\":[\"reasoning.encrypted_content\"],\"reasoning\":{\"effort\":\"medium\",\"summary\":\"auto\"},\"text\":{\"verbosity\":\"low\"}}"
+        },
+        {
+          "direction": "server",
+          "kind": "text",
+          "body": "{\"type\":\"response.created\",\"response\":{\"id\":\"resp_ws_reconnect_2\"}}"
+        },
+        {
+          "direction": "server",
+          "kind": "text",
+          "body": "{\"type\":\"response.output_item.added\",\"item\":{\"type\":\"message\",\"id\":\"msg_ws_reconnect_2\",\"role\":\"assistant\",\"content\":[]}}"
+        },
+        {
+          "direction": "server",
+          "kind": "text",
+          "body": "{\"type\":\"response.output_text.delta\",\"item_id\":\"msg_ws_reconnect_2\",\"delta\":\"Beta.\"}"
+        },
+        {
+          "direction": "server",
+          "kind": "text",
+          "body": "{\"type\":\"response.output_text.done\",\"item_id\":\"msg_ws_reconnect_2\",\"text\":\"Beta.\"}"
+        },
+        {
+          "direction": "server",
+          "kind": "text",
+          "body": "{\"type\":\"response.output_item.done\",\"item\":{\"type\":\"message\",\"id\":\"msg_ws_reconnect_2\",\"role\":\"assistant\",\"content\":[{\"type\":\"output_text\",\"text\":\"Beta.\"}]}}"
+        },
+        {
+          "direction": "server",
+          "kind": "text",
+          "body": "{\"type\":\"response.completed\",\"response\":{\"id\":\"resp_ws_reconnect_2\"}}"
+        }
+      ]
+    }
+  ]
+}

packages/ai/test/fixtures/recordings/openai-responses-websocket/recovers-from-explicit-continuation-rejection.json

@@ -0,0 +1,129 @@
+{
+  "version": 1,
+  "metadata": {
+    "provider": "openai",
+    "protocol": "openai-responses",
+    "transport": "websocket",
+    "model": "gpt-5.5",
+    "tags": [
+      "prefix:openai-responses-websocket",
+      "provider:openai",
+      "protocol:openai-responses",
+      "transport:websocket",
+      "continuation",
+      "recovery"
+    ],
+    "name": "openai-responses-websocket/recovers-from-explicit-continuation-rejection",
+    "recordedAt": "2026-08-20T00:00:00.000Z"
+  },
+  "interactions": [
+    {
+      "transport": "websocket",
+      "connection": {
+        "sequence": 0,
+        "url": "wss://api.openai.com/v1/responses",
+        "protocols": [],
+        "close": {
+          "code": 1000,
+          "reason": ""
+        }
+      },
+      "events": [
+        {
+          "direction": "client",
+          "kind": "text",
+          "body": "{\"type\":\"response.create\",\"model\":\"gpt-5.5\",\"input\":[{\"role\":\"system\",\"content\":\"Follow the user's exact reply instruction.\"},{\"role\":\"user\",\"content\":[{\"type\":\"input_text\",\"text\":\"Reply exactly: Ready.\"}]}],\"store\":false,\"max_output_tokens\":30,\"include\":[\"reasoning.encrypted_content\"],\"reasoning\":{\"effort\":\"medium\",\"summary\":\"auto\"},\"text\":{\"verbosity\":\"low\"}}"
+        },
+        {
+          "direction": "server",
+          "kind": "text",
+          "body": "{\"type\":\"response.created\",\"response\":{\"id\":\"resp_ws_rejection_1\"}}"
+        },
+        {
+          "direction": "server",
+          "kind": "text",
+          "body": "{\"type\":\"response.output_item.added\",\"item\":{\"type\":\"message\",\"id\":\"msg_ws_rejection_1\",\"role\":\"assistant\",\"content\":[]}}"
+        },
+        {
+          "direction": "server",
+          "kind": "text",
+          "body": "{\"type\":\"response.output_text.delta\",\"item_id\":\"msg_ws_rejection_1\",\"delta\":\"Ready.\"}"
+        },
+        {
+          "direction": "server",
+          "kind": "text",
+          "body": "{\"type\":\"response.output_text.done\",\"item_id\":\"msg_ws_rejection_1\",\"text\":\"Ready.\"}"
+        },
+        {
+          "direction": "server",
+          "kind": "text",
+          "body": "{\"type\":\"response.output_item.done\",\"item\":{\"type\":\"message\",\"id\":\"msg_ws_rejection_1\",\"role\":\"assistant\",\"content\":[{\"type\":\"output_text\",\"text\":\"Ready.\"}]}}"
+        },
+        {
+          "direction": "server",
+          "kind": "text",
+          "body": "{\"type\":\"response.completed\",\"response\":{\"id\":\"resp_ws_rejection_1\"}}"
+        }
+      ]
+    },
+    {
+      "transport": "websocket",
+      "connection": {
+        "sequence": 1,
+        "url": "wss://api.openai.com/v1/responses",
+        "protocols": [],
+        "close": {
+          "code": 1000,
+          "reason": ""
+        }
+      },
+      "events": [
+        {
+          "direction": "client",
+          "kind": "text",
+          "body": "{\"type\":\"response.create\",\"model\":\"gpt-5.5\",\"input\":[{\"role\":\"user\",\"content\":[{\"type\":\"input_text\",\"text\":\"Reply exactly: Recovered.\"}]}],\"store\":false,\"max_output_tokens\":30,\"previous_response_id\":\"resp_ws_rejection_1\",\"include\":[\"reasoning.encrypted_content\"],\"reasoning\":{\"effort\":\"medium\",\"summary\":\"auto\"},\"text\":{\"verbosity\":\"low\"}}"
+        },
+        {
+          "direction": "server",
+          "kind": "text",
+          "body": "{\"type\":\"error\",\"error\":{\"code\":\"previous_response_not_found\",\"message\":\"Previous response not found\"}}"
+        },
+        {
+          "direction": "client",
+          "kind": "text",
+          "body": "{\"type\":\"response.create\",\"model\":\"gpt-5.5\",\"input\":[{\"role\":\"system\",\"content\":\"Follow the user's exact reply instruction.\"},{\"role\":\"user\",\"content\":[{\"type\":\"input_text\",\"text\":\"Reply exactly: Ready.\"}]},{\"type\":\"message\",\"id\":\"msg_ws_rejection_1\",\"role\":\"assistant\",\"content\":[{\"type\":\"output_text\",\"text\":\"Ready.\"}]},{\"role\":\"user\",\"content\":[{\"type\":\"input_text\",\"text\":\"Reply exactly: Recovered.\"}]}],\"store\":false,\"max_output_tokens\":30,\"include\":[\"reasoning.encrypted_content\"],\"reasoning\":{\"effort\":\"medium\",\"summary\":\"auto\"},\"text\":{\"verbosity\":\"low\"}}"
+        },
+        {
+          "direction": "server",
+          "kind": "text",
+          "body": "{\"type\":\"response.created\",\"response\":{\"id\":\"resp_ws_rejection_2\"}}"
+        },
+        {
+          "direction": "server",
+          "kind": "text",
+          "body": "{\"type\":\"response.output_item.added\",\"item\":{\"type\":\"message\",\"id\":\"msg_ws_rejection_2\",\"role\":\"assistant\",\"content\":[]}}"
+        },
+        {
+          "direction": "server",
+          "kind": "text",
+          "body": "{\"type\":\"response.output_text.delta\",\"item_id\":\"msg_ws_rejection_2\",\"delta\":\"Recovered.\"}"
+        },
+        {
+          "direction": "server",
+          "kind": "text",
+          "body": "{\"type\":\"response.output_text.done\",\"item_id\":\"msg_ws_rejection_2\",\"text\":\"Recovered.\"}"
+        },
+        {
+          "direction": "server",
+          "kind": "text",
+          "body": "{\"type\":\"response.output_item.done\",\"item\":{\"type\":\"message\",\"id\":\"msg_ws_rejection_2\",\"role\":\"assistant\",\"content\":[{\"type\":\"output_text\",\"text\":\"Recovered.\"}]}}"
+        },
+        {
+          "direction": "server",
+          "kind": "text",
+          "body": "{\"type\":\"response.completed\",\"response\":{\"id\":\"resp_ws_rejection_2\"}}"
+        }
+      ]
+    }
+  ]
+}

packages/ai/test/llm.test.ts

@@ -121,7 +121,6 @@ describe("llm constructors", () => {
     const model = chatRoute.model({
       id: "kimi-k2",
       defaults: {
-        limits: { context: 128_000, output: 8_192 },
         generation: { maxTokens: 1_024, stop: ["END"] },
         providerOptions: { parallelToolCalls: false },
         http: { body: { extra_body: true } },
@@ -130,7 +129,6 @@ describe("llm constructors", () => {
     })
     const request = LLM.request({ model, prompt: "Say hello." })
 
-    expect(request.model.defaults?.limits).toEqual({ context: 128_000, output: 8_192 })
     expect(request.model.defaults?.generation).toEqual({ maxTokens: 1_024, stop: ["END"] })
     expect(request.model.defaults?.providerOptions).toEqual({ parallelToolCalls: false })
     expect(request.model.defaults?.http).toEqual({ body: { extra_body: true } })

packages/ai/test/provider-package.test.ts

@@ -43,7 +43,6 @@ describe("provider package entrypoints", () => {
       baseURL: "https://provider.example.test/v1",
       headers: { "x-application": "opencode" },
       body: { service_tier: "priority" },
-      limits: { context: 200_000, output: 64_000 },
     }
     const openrouter = OpenRouter.model("anthropic/claude-sonnet-4", {
       ...settings,
@@ -58,7 +57,6 @@ describe("provider package entrypoints", () => {
       expect(selected.route.endpoint.baseURL).toBe(settings.baseURL)
       expect(selected.route.defaults.headers).toEqual(settings.headers)
       expect(selected.route.defaults.http?.body).toEqual(settings.body)
-      expect(selected.route.defaults.limits).toEqual(settings.limits)
     }
     expect(openrouter.route.defaults.providerOptions).toEqual({ usage: true })
     expect(xai.route.defaults.providerOptions).toMatchObject({ reasoningEffort: "high", store: false })
@@ -70,14 +68,12 @@ describe("provider package entrypoints", () => {
       baseURL: "https://api.openai.test/v1",
       headers: { "x-application": "opencode" },
       body: { service_tier: "priority" },
-      limits: { context: 200_000, output: 64_000 },
       unrelatedInheritedSetting: true,
     })
 
     expect(selected.route.id).toBe("openai-responses")
     expect(selected.route.defaults.headers).toEqual({ "x-application": "opencode" })
     expect(selected.route.defaults.http?.body).toEqual({ service_tier: "priority" })
-    expect(selected.route.defaults.limits).toEqual({ context: 200_000, output: 64_000 })
   })
 
   test("maps OpenAI-compatible Responses settings onto the executable model", async () => {
@@ -88,7 +84,6 @@ describe("provider package entrypoints", () => {
       provider: "example",
       headers: { "x-application": "opencode" },
       body: { service_tier: "priority" },
-      limits: { context: 200_000, output: 64_000 },
       providerOptions: { reasoningEffort: "low", store: true },
     })
 
@@ -100,7 +95,6 @@ describe("provider package entrypoints", () => {
     })
     expect(selected.route.defaults.headers).toEqual({ "x-application": "opencode" })
     expect(selected.route.defaults.http?.body).toEqual({ service_tier: "priority" })
-    expect(selected.route.defaults.limits).toEqual({ context: 200_000, output: 64_000 })
     expect(selected.route.defaults.providerOptions).toEqual({ reasoningEffort: "low", store: true })
   })
 
@@ -112,7 +106,6 @@ describe("provider package entrypoints", () => {
       provider: "example",
       headers: { "x-application": "opencode" },
       body: { metadata: { user_id: "user_1" } },
-      limits: { context: 200_000, output: 64_000 },
       providerOptions: { effort: "low" },
     })
 
@@ -124,7 +117,6 @@ describe("provider package entrypoints", () => {
     })
     expect(selected.route.defaults.headers).toEqual({ "x-application": "opencode" })
     expect(selected.route.defaults.http?.body).toEqual({ metadata: { user_id: "user_1" } })
-    expect(selected.route.defaults.limits).toEqual({ context: 200_000, output: 64_000 })
     expect(selected.route.defaults.providerOptions).toEqual({ effort: "low" })
   })
 
@@ -185,7 +177,6 @@ describe("provider package entrypoints", () => {
       resourceName: "opencode-test",
       headers: { "x-application": "opencode" },
       body: { service_tier: "priority" },
-      limits: { context: 200_000, output: 64_000 },
     }
 
     const responses = AzureResponses.model("deployment", settings)
@@ -196,7 +187,6 @@ describe("provider package entrypoints", () => {
     expect(responses.route.endpoint.baseURL).toBe("https://opencode-test.openai.azure.com/openai/v1")
     expect(responses.route.defaults.headers).toEqual({ "x-application": "opencode" })
     expect(responses.route.defaults.http?.body).toEqual({ service_tier: "priority" })
-    expect(responses.route.defaults.limits).toEqual({ context: 200_000, output: 64_000 })
     expect(chat.route.id).toBe("azure-openai-chat")
   })
 
@@ -228,7 +218,6 @@ describe("provider package entrypoints", () => {
       baseURL: "https://generativelanguage.test/v1beta",
       headers: { "x-application": "opencode" },
       body: { safetySettings: [] },
-      limits: { context: 1_000_000, output: 65_536 },
       providerOptions: { thinkingConfig: { thinkingBudget: 1_024 } },
     })
 
@@ -236,7 +225,6 @@ describe("provider package entrypoints", () => {
     expect(selected.route.endpoint.baseURL).toBe("https://generativelanguage.test/v1beta")
     expect(selected.route.defaults.headers).toEqual({ "x-application": "opencode" })
     expect(selected.route.defaults.http?.body).toEqual({ safetySettings: [] })
-    expect(selected.route.defaults.limits).toEqual({ context: 1_000_000, output: 65_536 })
     expect(selected.route.defaults.providerOptions).toEqual({ thinkingConfig: { thinkingBudget: 1_024 } })
   })
 
@@ -250,7 +238,6 @@ describe("provider package entrypoints", () => {
       apiKey: "fixture",
       headers: { "x-application": "opencode" },
       body: { safetySettings: [] },
-      limits: { context: 1_000_000, output: 65_536 },
     })
     const messages = GoogleVertexMessages.model("claude-sonnet-4-6", {
       accessToken: "fixture",
@@ -274,7 +261,6 @@ describe("provider package entrypoints", () => {
     expect(gemini.route.endpoint.baseURL).toBe("https://aiplatform.googleapis.com/v1/publishers/google")
     expect(gemini.route.defaults.headers).toEqual({ "x-application": "opencode" })
     expect(gemini.route.defaults.http?.body).toEqual({ safetySettings: [] })
-    expect(gemini.route.defaults.limits).toEqual({ context: 1_000_000, output: 65_536 })
     expect(
       GoogleVertex.model("gemini-3.5-flash", {
         accessToken: "fixture",

packages/ai/test/provider/anthropic-messages.test.ts

@@ -322,7 +322,7 @@ describe("Anthropic Messages route", () => {
           { role: "user", content: [{ type: "tool_result", tool_use_id: "call_1", content: '{"forecast":"sunny"}' }] },
         ],
         stream: true,
-        max_tokens: 4096,
+        max_tokens: 32_000,
       })
     }),
   )

packages/ai/test/provider/gemini.test.ts

@@ -139,6 +139,48 @@ describe("Gemini route", () => {
     }),
   )
 
+  it.effect("keeps system updates separate from function responses", () =>
+    Effect.gen(function* () {
+      const prepared = yield* compileRequest(
+        LLM.request({
+          model,
+          messages: [
+            Message.assistant([ToolCallPart.make({ id: "call_1", name: "lookup", input: { query: "weather" } })]),
+            Message.tool({ id: "call_1", name: "lookup", result: "done", resultType: "text" }),
+            Message.system("Update."),
+            Message.system("Later update."),
+          ],
+        }),
+      )
+
+      expect(prepared.body.contents).toEqual([
+        {
+          role: "model",
+          parts: [{ functionCall: { id: undefined, name: "lookup", args: { query: "weather" } } }],
+        },
+        {
+          role: "user",
+          parts: [
+            {
+              functionResponse: {
+                id: undefined,
+                name: "lookup",
+                response: { name: "lookup", content: "done" },
+              },
+            },
+          ],
+        },
+        {
+          role: "user",
+          parts: [
+            { text: "<system-update>\nUpdate.\n</system-update>" },
+            { text: "<system-update>\nLater update.\n</system-update>" },
+          ],
+        },
+      ])
+    }),
+  )
+
   it.effect("prepares multimodal user input and tool history", () =>
     Effect.gen(function* () {
       const prepared = yield* compileRequest(

packages/ai/test/provider/openai-chat.test.ts

@@ -664,6 +664,74 @@ describe("OpenAI Chat route", () => {
     }),
   )
 
+  it.effect("preserves streamed refusals as ordinary assistant text", () =>
+    Effect.gen(function* () {
+      const response = yield* LLMClient.generate(request).pipe(
+        Effect.provide(
+          fixedResponse(
+            sseEvents(
+              deltaChunk({ role: "assistant", refusal: "I can't" }),
+              deltaChunk({ refusal: " help with that." }),
+              deltaChunk({}, "stop"),
+            ),
+          ),
+        ),
+      )
+
+      expect(response.text).toBe("I can't help with that.")
+      expect(response.finishReason).toEqual({ normalized: "stop", raw: "stop" })
+      expect(response.message.content).toEqual([{ type: "text", text: "I can't help with that." }])
+
+      const replay = yield* compileRequest(LLM.request({ model, messages: [response.message] }))
+      expect(replay.body.messages).toEqual([{ role: "assistant", content: "I can't help with that." }])
+    }),
+  )
+
+  it.effect("orders metadata-only reasoning before refusal output", () =>
+    Effect.gen(function* () {
+      const response = yield* LLMClient.generate(request).pipe(
+        Effect.provide(
+          fixedResponse(
+            sseEvents(
+              { choices: [{ delta: { reasoning_details: [] } }] },
+              deltaChunk({ refusal: "I can't help with that." }),
+              deltaChunk({}, "stop"),
+            ),
+          ),
+        ),
+      )
+
+      expect(response.message.content).toEqual([
+        { type: "reasoning", text: "", providerMetadata: { openai: { reasoningDetails: [] } } },
+        {
+          type: "text",
+          text: "I can't help with that.",
+        },
+      ])
+    }),
+  )
+
+  it.effect("joins content and refusal deltas into ordinary assistant text", () =>
+    Effect.gen(function* () {
+      const response = yield* LLMClient.generate(request).pipe(
+        Effect.provide(
+          fixedResponse(
+            sseEvents(
+              deltaChunk({ refusal: "No." }),
+              deltaChunk({ content: " Alternative." }),
+              deltaChunk({ refusal: " Still no." }),
+              deltaChunk({}, "stop"),
+            ),
+          ),
+        ),
+      )
+
+      expect(response.text).toBe("No. Alternative. Still no.")
+      expect(response.events.filter(LLMEvent.is.textStart).map((event) => event.id)).toEqual(["text-0"])
+      expect(response.events.filter(LLMEvent.is.textEnd).map((event) => event.id)).toEqual(["text-0"])
+    }),
+  )
+
   it.effect("parses and replays OpenAI-compatible reasoning fields", () =>
     Effect.gen(function* () {
       const fields = ["reasoning_content", "reasoning", "reasoning_text"] as const

packages/ai/test/provider/openai-compatible-responses.test.ts

@@ -73,7 +73,7 @@ describe("Open Responses-compatible route", () => {
       expect(prepared.body.input).toEqual([
         { role: "user", content: [{ type: "input_text", text: "Before." }] },
         { role: "developer", content: "Operator update." },
-        { role: "assistant", content: [{ type: "output_text", text: "After." }] },
+        { type: "message", role: "assistant", content: [{ type: "output_text", text: "After." }] },
       ])
     }),
   )
@@ -113,11 +113,67 @@ describe("Open Responses-compatible route", () => {
       )
 
       expect(prepared.body).toMatchObject({
-        input: [{ role: "assistant", content: [{ type: "output_text", text: "Unclassified." }] }],
+        input: [{ type: "message", role: "assistant", content: [{ type: "output_text", text: "Unclassified." }] }],
       })
     }),
   )
 
+  it.effect("preserves standard refusal content as ordinary assistant text", () =>
+    Effect.gen(function* () {
+      const model = configure({
+        apiKey: "test-key",
+        baseURL: "https://responses.example.test/v1",
+        provider: "example",
+      }).model("example-model")
+      const response = yield* LLMClient.generate(LLM.request({ model, prompt: "Unsafe request" })).pipe(
+        Effect.provide(
+          fixedResponse(
+            sseEvents(
+              {
+                type: "response.output_item.added",
+                output_index: 0,
+                item: { type: "message", id: "msg_refusal", content: [] },
+              },
+              {
+                type: "response.refusal.done",
+                item_id: "msg_refusal",
+                refusal: "I can't help with that.",
+              },
+              {
+                type: "response.output_item.done",
+                output_index: 0,
+                item: {
+                  type: "message",
+                  id: "msg_refusal",
+                  content: [{ type: "refusal", refusal: "I can't help with that." }],
+                },
+              },
+              { type: "response.completed", response: { id: "resp_1" } },
+            ),
+          ),
+        ),
+      )
+
+      expect(response.message.content).toEqual([
+        {
+          type: "text",
+          text: "I can't help with that.",
+          providerMetadata: { openresponses: { itemId: "msg_refusal" } },
+        },
+      ])
+
+      const prepared = yield* compileRequest(LLM.request({ model, messages: [response.message] }))
+      expect(prepared.body.input).toEqual([
+        {
+          type: "message",
+          id: "msg_refusal",
+          role: "assistant",
+          content: [{ type: "output_text", text: "I can't help with that." }],
+        },
+      ])
+    }),
+  )
+
   it.effect("reads standard Open Responses options", () =>
     Effect.gen(function* () {
       const model = configure({
@@ -126,6 +182,10 @@ describe("Open Responses-compatible route", () => {
         providerOptions: {
           reasoningEffort: "low",
           store: true,
+          metadata: { environment: "test" },
+          safetyIdentifier: "user_123",
+          streamOptions: { includeObfuscation: false },
+          topLogprobs: 3,
           truncation: "auto",
           allowedTools: { toolNames: ["lookup"] },
           maxToolCalls: 2,
@@ -136,6 +196,7 @@ describe("Open Responses-compatible route", () => {
         LLM.request({
           model,
           prompt: "Think.",
+          generation: { presencePenalty: 0.2, frequencyPenalty: -0.1 },
           tools: [ToolDefinition.make({ name: "lookup", description: "Lookup data", inputSchema: { type: "object" } })],
         }),
       )
@@ -143,6 +204,12 @@ describe("Open Responses-compatible route", () => {
       expect(prepared.body).toMatchObject({
         reasoning: { effort: "low" },
         store: true,
+        metadata: { environment: "test" },
+        safety_identifier: "user_123",
+        stream_options: { include_obfuscation: false },
+        top_logprobs: 3,
+        presence_penalty: 0.2,
+        frequency_penalty: -0.1,
         truncation: "auto",
         tool_choice: {
           type: "allowed_tools",

packages/ai/test/provider/openai-responses-websocket.recorded.test.ts

@@ -0,0 +1,216 @@
+import { describe, expect } from "bun:test"
+import { Effect, Stream } from "effect"
+import { Socket } from "effect/unstable/socket"
+import { LLM, LLMRequest, Message, ToolRuntime } from "../../src/index.js"
+import {
+  LLMClient,
+  WebSocketTransport,
+  type ChannelCheckpoint,
+  type ChannelObservation,
+  type WebSocketChannelExchange,
+  type WebSocketChannelExecutor,
+  type WebSocketConnection,
+} from "../../src/route.js"
+import { configure } from "../../src/providers/openai.js"
+import { decodeJson } from "../../src/protocols/shared.js"
+import { weatherRuntimeTool, weatherTool, weatherToolName } from "../recorded-scenarios.js"
+import { recordedTests } from "../recorded-test.js"
+
+const model = configure({ apiKey: process.env.OPENAI_API_KEY ?? "fixture" }).responses("gpt-5.5")
+const recorded = recordedTests({
+  prefix: "openai-responses-websocket",
+  provider: "openai",
+  protocol: "openai-responses",
+  requires: ["OPENAI_API_KEY"],
+  tags: ["transport:websocket"],
+  metadata: { transport: "websocket", model: model.id },
+})
+
+const observationFrame = (observation: ChannelObservation) => {
+  if (observation.type === "frame" || observation.type === "completed" || observation.type === "incomplete")
+    return Effect.succeed(observation.frame)
+  return Effect.fail(observation.error)
+}
+
+const terminal = (observation: ChannelObservation) => observation.type !== "frame"
+
+// This deliberately models only sequential test traffic. Core owns production connection pooling and recovery.
+const makeChannel = Effect.gen(function* () {
+  const constructor = yield* Socket.WebSocketConstructor
+  let connection: WebSocketConnection | undefined
+  let checkpoint: ChannelCheckpoint | undefined
+  let pending: ChannelCheckpoint | undefined
+  let opens = 0
+  const sent: unknown[] = []
+
+  const close = Effect.suspend(() => {
+    const current = connection
+    connection = undefined
+    return current ? current.close : Effect.void
+  })
+  yield* Effect.addFinalizer(() => close)
+
+  const executor: WebSocketChannelExecutor = {
+    execute: (exchange: WebSocketChannelExchange) =>
+      Effect.gen(function* () {
+        if (!connection) {
+          connection = yield* WebSocketTransport.open(exchange.connect).pipe(
+            Effect.provideService(Socket.WebSocketConstructor, constructor),
+          )
+          opens += 1
+        }
+        const current = connection
+        const create = yield* exchange.driver.create(checkpoint)
+        if (create.mode === "full") checkpoint = undefined
+        pending = undefined
+        sent.push(decodeJson(create.message))
+        yield* current.sendText(create.message)
+        const decoder = new TextDecoder()
+        return {
+          frames: current.messages.pipe(
+            Stream.map((message) => WebSocketTransport.messageText(message, decoder)),
+            Stream.mapEffect((frame) => exchange.driver.observe(create, frame)),
+            Stream.tap((observation) =>
+              Effect.sync(() => {
+                if (!terminal(observation)) return
+                pending = observation.type === "completed" ? observation.checkpoint : undefined
+                if (observation.type !== "completed") checkpoint = undefined
+              }),
+            ),
+            Stream.takeUntil(terminal),
+            Stream.mapEffect(observationFrame),
+          ),
+          complete: Effect.sync(() => {
+            checkpoint = pending
+            pending = undefined
+          }),
+        }
+      }),
+  }
+
+  return {
+    executor,
+    sent,
+    opens: () => opens,
+    reconnect: (preserveCheckpoint = false) =>
+      close.pipe(
+        Effect.andThen(
+          Effect.sync(() => {
+            pending = undefined
+            if (!preserveCheckpoint) checkpoint = undefined
+          }),
+        ),
+      ),
+  }
+})
+
+describe("OpenAI Responses WebSocket recorded", () => {
+  recorded.effect.with("continues a tool call over one socket", { tags: ["tool", "continuation"] }, () =>
+    Effect.gen(function* () {
+      const channel = yield* makeChannel
+      const request = LLM.request({
+        id: "recorded_openai_responses_websocket_tool",
+        model,
+        system: "Call get_weather once, then reply exactly: Paris is sunny.",
+        prompt: "What is the weather in Paris?",
+        tools: [weatherTool],
+        generation: { maxTokens: 50 },
+        cache: "none",
+      })
+      const first = yield* LLMClient.generate(request, { webSocket: channel.executor })
+      const call = first.toolCalls[0]
+      if (!call) yield* Effect.die("Expected get_weather tool call")
+      const result = yield* ToolRuntime.dispatch({ [weatherToolName]: weatherRuntimeTool }, call)
+      const second = yield* LLMClient.generate(
+        LLMRequest.update(request, {
+          messages: [
+            ...request.messages,
+            first.message,
+            Message.tool({ id: call.id, name: call.name, result: result.result }),
+          ],
+        }),
+        { webSocket: channel.executor },
+      )
+
+      expect(second.text).toBe("Paris is sunny.")
+      expect(channel.opens()).toBe(1)
+      expect(channel.sent).toHaveLength(2)
+      expect(channel.sent[1]).toMatchObject({
+        previous_response_id: expect.any(String),
+        input: [{ type: "function_call_output", call_id: call.id, output: expect.any(String) }],
+      })
+    }),
+  )
+
+  recorded.effect.with("reconstructs full context after reconnect", { tags: ["reconnect", "full-context"] }, () =>
+    Effect.gen(function* () {
+      const channel = yield* makeChannel
+      const request = LLM.request({
+        id: "recorded_openai_responses_websocket_reconnect",
+        model,
+        system: "Follow the user's exact reply instruction.",
+        prompt: "Reply exactly: Alpha.",
+        generation: { maxTokens: 30 },
+        cache: "none",
+      })
+      const first = yield* LLMClient.generate(request, { webSocket: channel.executor })
+      yield* channel.reconnect()
+      const second = yield* LLMClient.generate(
+        LLMRequest.update(request, {
+          messages: [...request.messages, first.message, Message.user("Reply exactly: Beta.")],
+        }),
+        { webSocket: channel.executor },
+      )
+
+      expect(first.text).toBe("Alpha.")
+      expect(second.text).toBe("Beta.")
+      expect(channel.opens()).toBe(2)
+      expect(channel.sent[1]).not.toHaveProperty("previous_response_id")
+      expect(channel.sent[1]).toMatchObject({
+        input: [
+          { role: "system", content: "Follow the user's exact reply instruction." },
+          { role: "user", content: [{ type: "input_text", text: "Reply exactly: Alpha." }] },
+          { role: "assistant", content: [{ type: "output_text", text: "Alpha." }] },
+          { role: "user", content: [{ type: "input_text", text: "Reply exactly: Beta." }] },
+        ],
+      })
+    }),
+  )
+
+  recorded.effect.with("recovers from explicit continuation rejection", { tags: ["continuation", "recovery"] }, () =>
+    Effect.gen(function* () {
+      const channel = yield* makeChannel
+      const request = LLM.request({
+        id: "recorded_openai_responses_websocket_rejection",
+        model,
+        system: "Follow the user's exact reply instruction.",
+        prompt: "Reply exactly: Ready.",
+        generation: { maxTokens: 30 },
+        cache: "none",
+      })
+      const first = yield* LLMClient.generate(request, { webSocket: channel.executor })
+      const continuation = LLMRequest.update(request, {
+        messages: [...request.messages, first.message, Message.user("Reply exactly: Recovered.")],
+      })
+      yield* channel.reconnect(true)
+      const rejected = yield* LLMClient.generate(continuation, { webSocket: channel.executor }).pipe(Effect.flip)
+      const recovered = yield* LLMClient.generate(continuation, { webSocket: channel.executor })
+
+      expect(rejected).toMatchObject({
+        reason: { _tag: "Transport", delivery: "rejected", recovery: "retry-full" },
+      })
+      expect(recovered.text).toBe("Recovered.")
+      expect(channel.opens()).toBe(2)
+      expect(channel.sent[1]).toHaveProperty("previous_response_id", expect.any(String))
+      expect(channel.sent[2]).not.toHaveProperty("previous_response_id")
+      expect(channel.sent[2]).toMatchObject({
+        input: [
+          { role: "system", content: "Follow the user's exact reply instruction." },
+          { role: "user", content: [{ type: "input_text", text: "Reply exactly: Ready." }] },
+          { role: "assistant", content: [{ type: "output_text", text: "Ready." }] },
+          { role: "user", content: [{ type: "input_text", text: "Reply exactly: Recovered." }] },
+        ],
+      })
+    }),
+  )
+})

packages/ai/test/provider/openai-responses.test.ts

@@ -263,7 +263,7 @@ describe("OpenAI Responses route", () => {
       expect(prepared.body.input).toEqual([
         { role: "user", content: [{ type: "input_text", text: "Before." }] },
         { role: "developer", content: "Operator update." },
-        { role: "assistant", content: [{ type: "output_text", text: "After." }] },
+        { type: "message", role: "assistant", content: [{ type: "output_text", text: "After." }] },
       ])
     }),
   )
@@ -485,7 +485,7 @@ describe("OpenAI Responses route", () => {
     }),
   )
 
-  it.effect("continues store-false reasoning without replaying the output-only item ID", () =>
+  it.effect("continues store-false reasoning while retaining the output item ID", () =>
     Effect.gen(function* () {
       const firstInput = [{ role: "user", content: [{ type: "input_text", text: "Think" }] }]
       const request = { type: "response.create", model: "gpt-5.2", store: false, input: firstInput }
@@ -515,6 +515,7 @@ describe("OpenAI Responses route", () => {
           ...firstInput,
           {
             type: "reasoning",
+            id: "rs_1",
             summary: [{ type: "summary_text", text: "Thought" }],
             encrypted_content: "encrypted",
           },
@@ -1284,6 +1285,7 @@ describe("OpenAI Responses route", () => {
           model: OpenAI.configure({ baseURL: "https://api.openai.test/v1/", apiKey: "test" }).model("gpt-5.2"),
           prompt: "think",
           promptCacheKey: "session_123",
+          generation: { presencePenalty: 0.25, frequencyPenalty: -0.25 },
           tools: [
             ToolDefinition.make({ name: "read", description: "Read a file", inputSchema: { type: "object" } }),
             ToolDefinition.make({ name: "grep", description: "Search files", inputSchema: { type: "object" } }),
@@ -1293,6 +1295,10 @@ describe("OpenAI Responses route", () => {
             reasoningEffort: "high",
             reasoningSummary: "auto",
             include: ["reasoning.encrypted_content"],
+            metadata: { environment: "test", tenant: "acme" },
+            safetyIdentifier: "user_123",
+            streamOptions: { includeObfuscation: false },
+            topLogprobs: 5,
             truncation: "disabled",
             allowedTools: { toolNames: ["read", "grep"], mode: "required" },
             maxToolCalls: 4,
@@ -1306,6 +1312,12 @@ describe("OpenAI Responses route", () => {
       expect(prepared.body.include).toEqual(["reasoning.encrypted_content"])
       expect(prepared.body.reasoning).toEqual({ effort: "high", summary: "auto" })
       expect(prepared.body.text).toEqual({ verbosity: "low" })
+      expect(prepared.body.metadata).toEqual({ environment: "test", tenant: "acme" })
+      expect(prepared.body.safety_identifier).toBe("user_123")
+      expect(prepared.body.stream_options).toEqual({ include_obfuscation: false })
+      expect(prepared.body.top_logprobs).toBe(5)
+      expect(prepared.body.presence_penalty).toBe(0.25)
+      expect(prepared.body.frequency_penalty).toBe(-0.25)
       expect(prepared.body.truncation).toBe("disabled")
       expect(prepared.body.tool_choice).toEqual({
         type: "allowed_tools",
@@ -1473,7 +1485,7 @@ describe("OpenAI Responses route", () => {
       expect(response.text).toBe("Hello!")
       expect(response.events).toEqual([
         { type: "step-start", index: 0 },
-        { type: "text-start", id: "msg_1" },
+        { type: "text-start", id: "msg_1", providerMetadata: { openai: { itemId: "msg_1" } } },
         { type: "text-delta", id: "msg_1", text: "Hello" },
         { type: "text-delta", id: "msg_1", text: "!" },
         { type: "text-end", id: "msg_1" },
@@ -1494,6 +1506,108 @@ describe("OpenAI Responses route", () => {
     }),
   )
 
+  it.effect("preserves standard refusal content as ordinary assistant text", () =>
+    Effect.gen(function* () {
+      const response = yield* LLMClient.generate(request).pipe(
+        Effect.provide(
+          fixedResponse(
+            sseEvents(
+              {
+                type: "response.output_item.added",
+                output_index: 0,
+                item: { type: "message", id: "msg_refusal", content: [] },
+              },
+              {
+                type: "response.content_part.added",
+                item_id: "msg_refusal",
+                output_index: 0,
+                content_index: 0,
+                part: { type: "refusal", refusal: "" },
+              },
+              {
+                type: "response.refusal.delta",
+                item_id: "msg_refusal",
+                output_index: 0,
+                content_index: 0,
+                delta: "I can't",
+              },
+              {
+                type: "response.refusal.delta",
+                item_id: "msg_refusal",
+                output_index: 0,
+                content_index: 0,
+                delta: " help with that.",
+              },
+              {
+                type: "response.refusal.done",
+                item_id: "msg_refusal",
+                output_index: 0,
+                content_index: 0,
+                refusal: "I can't help with that.",
+              },
+              {
+                type: "response.content_part.done",
+                item_id: "msg_refusal",
+                output_index: 0,
+                content_index: 0,
+                part: { type: "refusal", refusal: "I can't help with that." },
+              },
+              {
+                type: "response.output_item.done",
+                output_index: 0,
+                item: {
+                  type: "message",
+                  id: "msg_refusal",
+                  phase: "final_answer",
+                  content: [{ type: "refusal", refusal: "I can't help with that." }],
+                },
+              },
+              { type: "response.completed", response: { id: "resp_1" } },
+            ),
+          ),
+        ),
+      )
+
+      expect(response.text).toBe("I can't help with that.")
+      expect(response.finishReason).toEqual({ normalized: "stop", raw: undefined })
+      expect(response.message.content).toEqual([
+        {
+          type: "text",
+          text: "I can't help with that.",
+          providerMetadata: { openai: { itemId: "msg_refusal", phase: "final_answer" } },
+        },
+      ])
+
+      const prepared = yield* compileRequest(LLM.request({ model, messages: [response.message] }))
+      expect(prepared.body.input).toEqual([
+        {
+          type: "message",
+          id: "msg_refusal",
+          role: "assistant",
+          content: [{ type: "output_text", text: "I can't help with that." }],
+          phase: "final_answer",
+        },
+      ])
+    }),
+  )
+
+  it.effect("rejects malformed refusal events", () =>
+    Effect.gen(function* () {
+      const events = [
+        { type: "response.refusal.delta", output_index: 0, content_index: 0, delta: "missing item" },
+        { type: "response.refusal.delta", item_id: "msg_1", output_index: 0, content_index: 0 },
+        { type: "response.refusal.done", item_id: "msg_1", output_index: 0, content_index: 0 },
+      ]
+      for (const event of events) {
+        const error = yield* LLMClient.generate(request).pipe(
+          Effect.provide(fixedResponse(sseEvents(event))),
+          Effect.flip,
+        )
+        expect(error.reason._tag).toBe("InvalidProviderOutput")
+      }
+    }),
+  )
+
   it.effect("preserves and replays assistant message phases", () =>
     Effect.gen(function* () {
       const response = yield* LLMClient.generate(request).pipe(
@@ -1532,33 +1646,39 @@ describe("OpenAI Responses route", () => {
         {
           type: "text",
           text: "Checking.",
-          providerMetadata: { openai: { phase: "commentary" } },
+          providerMetadata: { openai: { itemId: "msg_commentary", phase: "commentary" } },
         },
         {
           type: "text",
           text: "Finished.",
-          providerMetadata: { openai: { phase: "final_answer" } },
+          providerMetadata: { openai: { itemId: "msg_final", phase: "final_answer" } },
         },
         {
           type: "text",
           text: "Unclassified.",
-          providerMetadata: { openai: { phase: null } },
+          providerMetadata: { openai: { itemId: "msg_null", phase: null } },
         },
       ])
 
       const prepared = yield* compileRequest(LLM.request({ model, messages: [response.message] }))
       expect(prepared.body.input).toEqual([
         {
+          type: "message",
+          id: "msg_commentary",
           role: "assistant",
           content: [{ type: "output_text", text: "Checking." }],
           phase: "commentary",
         },
         {
+          type: "message",
+          id: "msg_final",
           role: "assistant",
           content: [{ type: "output_text", text: "Finished." }],
           phase: "final_answer",
         },
         {
+          type: "message",
+          id: "msg_null",
           role: "assistant",
           content: [{ type: "output_text", text: "Unclassified." }],
           phase: null,
@@ -1652,12 +1772,12 @@ describe("OpenAI Responses route", () => {
       )
 
       expect(response.events.filter((event) => event.type.startsWith("text-"))).toEqual([
-        { type: "text-start", id: "msg_1" },
+        { type: "text-start", id: "msg_1", providerMetadata: { openai: { itemId: "msg_1" } } },
         { type: "text-delta", id: "msg_1", text: "First" },
-        { type: "text-end", id: "msg_1" },
-        { type: "text-start", id: "msg_2" },
+        { type: "text-end", id: "msg_1", providerMetadata: undefined },
+        { type: "text-start", id: "msg_2", providerMetadata: { openai: { itemId: "msg_2" } } },
         { type: "text-delta", id: "msg_2", text: "Second" },
-        { type: "text-end", id: "msg_2" },
+        { type: "text-end", id: "msg_2", providerMetadata: { openai: { itemId: "msg_2" } } },
       ])
     }),
   )
@@ -1689,7 +1809,7 @@ describe("OpenAI Responses route", () => {
       expect(response.events.filter((event) => event.type === "finish")).toHaveLength(1)
       expect(response.message.content).toEqual([
         { type: "reasoning", text: "thinking" },
-        { type: "text", text: "Hello" },
+        { type: "text", text: "Hello", providerMetadata: { openai: { itemId: "msg_1" } } },
       ])
     }),
   )
@@ -1850,6 +1970,7 @@ describe("OpenAI Responses route", () => {
                   { role: "user", content: [{ type: "input_text", text: "What changed?" }] },
                   {
                     type: "reasoning",
+                    id: "rs_1",
                     encrypted_content: "encrypted-state",
                     summary: [{ type: "summary_text", text: "Checked the previous diff." }],
                   },
@@ -1857,7 +1978,6 @@ describe("OpenAI Responses route", () => {
                   { role: "user", content: [{ type: "input_text", text: "Summarize it." }] },
                 ],
               })
-              expect(body.input[1]).not.toHaveProperty("id")
               return input.respond(
                 sseEvents(
                   { type: "response.output_text.delta", item_id: "msg_1", delta: "Parser now round-trips reasoning." },
@@ -1901,13 +2021,14 @@ describe("OpenAI Responses route", () => {
       )
 
       expect(prepared.body.input).toEqual([
-        { role: "assistant", content: [{ type: "output_text", text: "Before." }] },
+        { type: "message", role: "assistant", content: [{ type: "output_text", text: "Before." }] },
         {
           type: "reasoning",
+          id: "rs_1",
           encrypted_content: "encrypted-state",
           summary: [{ type: "summary_text", text: "Checked order." }],
         },
-        { role: "assistant", content: [{ type: "output_text", text: "After." }] },
+        { type: "message", role: "assistant", content: [{ type: "output_text", text: "After." }] },
       ])
     }),
   )
@@ -2039,6 +2160,7 @@ describe("OpenAI Responses route", () => {
       expect(prepared.body.input).toEqual([
         {
           type: "reasoning",
+          id: "rs_1",
           encrypted_content: "encrypted-state",
           summary: [
             { type: "summary_text", text: "First" },
@@ -2171,6 +2293,50 @@ describe("OpenAI Responses route", () => {
           usage,
         },
       ])
+
+      const prepared = yield* compileRequest(LLM.request({ model, messages: [response.message] }))
+      expect(prepared.body.input).toEqual([
+        {
+          type: "function_call",
+          id: "item_1",
+          call_id: "call_1",
+          name: "lookup",
+          arguments: '{"query":"weather"}',
+        },
+      ])
+    }),
+  )
+
+  it.effect("finalizes a pending function call at response completion", () =>
+    Effect.gen(function* () {
+      const body = sseEvents(
+        {
+          type: "response.output_item.added",
+          item: { type: "function_call", id: "item_1", call_id: "call_1", name: "lookup", arguments: "" },
+        },
+        { type: "response.completed", response: { usage: { input_tokens: 5, output_tokens: 1 } } },
+      )
+      const response = yield* LLMClient.generate(request).pipe(Effect.provide(fixedResponse(body)))
+
+      expect(response.events.filter((event) => LLMEvent.is.toolInputEnd(event) || LLMEvent.is.toolCall(event))).toEqual(
+        [
+          {
+            type: "tool-input-end",
+            id: "call_1",
+            name: "lookup",
+            providerMetadata: { openai: { itemId: "item_1" } },
+          },
+          {
+            type: "tool-call",
+            id: "call_1",
+            name: "lookup",
+            input: {},
+            providerExecuted: undefined,
+            providerMetadata: { openai: { itemId: "item_1" } },
+          },
+        ],
+      )
+      expect(response.finishReason.normalized).toBe("tool-calls")
     }),
   )
 
@@ -2237,6 +2403,35 @@ describe("OpenAI Responses route", () => {
     }),
   )
 
+  it.effect("retains function call item metadata when output_item.added is absent", () =>
+    Effect.gen(function* () {
+      const response = yield* LLMClient.generate(request).pipe(
+        Effect.provide(
+          fixedResponse(
+            sseEvents(
+              {
+                type: "response.output_item.done",
+                item: {
+                  type: "function_call",
+                  id: "item_1",
+                  call_id: "call_1",
+                  name: "lookup",
+                  arguments: '{"query":"weather"}',
+                },
+              },
+              { type: "response.completed", response: { id: "resp_1" } },
+            ),
+          ),
+        ),
+      )
+
+      expect(response.events.find(LLMEvent.is.toolCall)).toMatchObject({
+        id: "call_1",
+        providerMetadata: { openai: { itemId: "item_1" } },
+      })
+    }),
+  )
+
   it.effect("decodes web_search_call as provider-executed tool-call + tool-result", () =>
     Effect.gen(function* () {
       const item = {

packages/ai/test/recorded-test.ts

@@ -1,5 +1,7 @@
 import { HttpRecorder } from "@opencode-ai/http-recorder"
+import { NodeSocket } from "@effect/platform-node"
 import { Layer } from "effect"
+import { Socket } from "effect/unstable/socket"
 import * as path from "node:path"
 import { fileURLToPath } from "node:url"
 import { LLMClient, RequestExecutor } from "../src/route.js"
@@ -16,7 +18,7 @@ import {
 const __dirname = path.dirname(fileURLToPath(import.meta.url))
 const FIXTURES_DIR = path.resolve(__dirname, "fixtures", "recordings")
 
-type RecordedEnv = RequestExecutorService | LLMClientService | ImageClientService
+type RecordedEnv = RequestExecutorService | LLMClientService | ImageClientService | Socket.WebSocketConstructor
 
 type RecordedTestsOptions = RecordedGroupOptions & {
   readonly options?: HttpRecorder.RecorderOptions
@@ -69,7 +71,7 @@ export const recordedTests = (options: RecordedTestsOptions) =>
         ...metadata,
       }
       if (recording) {
-        if (process.env.CI !== undefined) throw new Error("Unset CI before recording HTTP cassettes")
+        if (process.env.CI !== undefined) throw new Error("Unset CI before recording cassettes")
         HttpRecorder.removeCassetteSync(cassette, { directory: FIXTURES_DIR })
       }
       const requestExecutor = RequestExecutor.layer.pipe(
@@ -81,10 +83,16 @@ export const recordedTests = (options: RecordedTestsOptions) =>
           }),
         ),
       )
+      const webSocket = HttpRecorder.layerWebSocketConstructor(cassette, {
+        ...recorderOptions,
+        directory: FIXTURES_DIR,
+        metadata: recorderMetadata,
+      }).pipe(Layer.provide(NodeSocket.layerWebSocketConstructorWS))
       return Layer.mergeAll(
         requestExecutor,
         LLMClient.layer.pipe(Layer.provide(requestExecutor)),
         ImageClient.layer.pipe(Layer.provide(requestExecutor)),
+        webSocket,
       )
     },
   })

packages/app/e2e/performance/README.md

@@ -2,12 +2,34 @@
 
 The app's high-volume performance diagnostics live under `packages/app/e2e/performance` and are excluded from normal local and CI Playwright discovery. The benchmark config builds the app and serves the production bundle before running scenarios serially.
 
+The `devex` category is the explicit exception to the production-build rule. It measures development commands from submission through a user-visible ready state and has its own Playwright configuration.
+
 Run the suite explicitly from `packages/app`:
 
 ```sh
 bun run test:bench
 ```
 
+Run the desktop development startup benchmark from the repository root:
+
+```sh
+bun run bench:devex
+```
+
+It runs five serial samples of the exact `bun dev:desktop` command. Each sample uses a fresh desktop profile, database, service configuration, service registration, and service process; the desktop selects an isolated ephemeral loopback endpoint. It removes desktop build output and the desktop Vite cache before every run; dependencies, Bun's package cache, and Electron remain installed. The harness stops only that sample's service; it does not stop or change the elected global OpenCode service. The measured endpoint is a visible Home page whose empty-state controls pass Playwright actionability checks. The command's Electron installation check remains inside the measured interval.
+
+Set `DESKTOP_STARTUP_RUNS` only for focused diagnostics:
+
+```sh
+DESKTOP_STARTUP_RUNS=1 bun run bench:devex
+```
+
+Set `OPENCODE_PERFORMANCE_TRACE_DIR` to capture the renderer's CDP trace from attachment through actionable Home:
+
+```sh
+DESKTOP_STARTUP_RUNS=1 OPENCODE_PERFORMANCE_TRACE_DIR=/tmp/opencode-desktop-traces bun run bench:devex
+```
+
 PowerShell:
 
 ```powershell

packages/app/e2e/performance/chrome-trace.ts

@@ -18,9 +18,9 @@ const categories = [
   "disabled-by-default-v8.cpu_profiler",
 ]
 
-export async function startChromeTrace(page: Page, name: string) {
+export async function startChromeTrace(page: Page, name: string): Promise<undefined | (() => Promise<string>)> {
   const directory = process.env.OPENCODE_PERFORMANCE_TRACE_DIR
-  if (!directory) return
+  if (!directory) return undefined
 
   const selectors = process.env.OPENCODE_PERFORMANCE_SELECTOR_TRACE === "1"
   const file = await prepareChromeTrace(directory, name, selectors)

packages/app/e2e/performance/devex/desktop-startup-benchmark.spec.ts

@@ -0,0 +1,26 @@
+import { benchmark } from "../benchmark"
+import {
+  desktopBenchmarkContext,
+  runDesktopStartup,
+  summarizeDesktopStartup,
+  type DesktopStartupSample,
+} from "./desktop-startup"
+
+benchmark.describe("devex: desktop startup", () => {
+  benchmark("opens a cold desktop on Home", async ({ report }, testInfo) => {
+    benchmark.setTimeout(15 * 60_000)
+    const runs = Number(process.env.DESKTOP_STARTUP_RUNS ?? 5)
+    if (!Number.isSafeInteger(runs) || runs < 1) throw new Error("DESKTOP_STARTUP_RUNS must be a positive integer")
+
+    const samples: DesktopStartupSample[] = []
+    const context = await desktopBenchmarkContext(runs)
+    for (let run = 1; run <= runs; run++) {
+      const sample = await runDesktopStartup(run, testInfo).catch((error) => {
+        report(samples.length ? { samples, summary: summarizeDesktopStartup(samples) } : { samples }, context)
+        throw error
+      })
+      samples.push(sample)
+    }
+    report({ samples, summary: summarizeDesktopStartup(samples) }, context)
+  })
+})

packages/app/e2e/performance/devex/desktop-startup.ts

@@ -0,0 +1,526 @@
+import { Service } from "@opencode-ai/client/service"
+import { chromium, expect, type Browser, type Page, type TestInfo } from "@playwright/test"
+import { spawn, spawnSync, type ChildProcess } from "node:child_process"
+import { mkdir, mkdtemp, readFile, rm, writeFile } from "node:fs/promises"
+import { tmpdir } from "node:os"
+import { join, resolve } from "node:path"
+import { startChromeTrace } from "../chrome-trace"
+
+const repository = resolve(import.meta.dirname, "../../../../..")
+const milestones = [
+  "bunRootScript",
+  "bunDesktopScript",
+  "desktopPrepared",
+  "mainBundleReady",
+  "preloadBundleReady",
+  "rendererDevServerReady",
+  "electronSpawnStarted",
+  "debugEndpointReady",
+  "electronStarted",
+  "serviceEnsureStarted",
+  "serviceSpawnRequested",
+  "serviceReady",
+  "backgroundLoadingReady",
+  "rendererViteConnected",
+  "rendererInitializationStarted",
+  "rendererInitializationReady",
+  "windowVisible",
+  "homeReady",
+] as const
+const phases = [
+  "desktopPreparation",
+  "viteMainBundle",
+  "vitePreloadBundle",
+  "rendererServerStartup",
+  "electronStartup",
+  "serviceSpawnWait",
+  "serviceProcessStartup",
+  "rendererStartup",
+  "visibleWindowToHome",
+] as const
+
+type Milestone = (typeof milestones)[number]
+type Phase = (typeof phases)[number]
+type ServiceInfo = { id: string; version: string; url: string; pid: number }
+
+export type DesktopStartupSample = {
+  run: number
+  commandToHomeReadyMs: number
+  milestonesMs: Record<Milestone, number>
+  phasesMs: Record<Phase, number>
+  service: Omit<ServiceInfo, "id">
+}
+
+export async function runDesktopStartup(run: number, testInfo: TestInfo) {
+  const profile = await createColdProfile()
+  const desktop = await Promise.resolve()
+    .then(() => startDesktop(profile))
+    .catch(async (error) => {
+      await rm(profile.root, { recursive: true, force: true, maxRetries: 5, retryDelay: 100 })
+      throw error
+    })
+  try {
+    const page = await desktop.open()
+    const stopTrace = await startChromeTrace(page, `desktop-startup-${run}`)
+    try {
+      await startThemeObservation(page)
+      await waitForHome(page, desktop.mark)
+      await requireStableTheme(page)
+      return await desktop.result(run)
+    } finally {
+      await stopTrace?.()
+    }
+  } finally {
+    await desktop.close(testInfo, run)
+  }
+}
+
+export async function desktopBenchmarkContext(runs: number) {
+  const pkg = JSON.parse(await readFile(join(repository, "packages/desktop/package.json"), "utf8"))
+  const revision = spawnSync("git", ["rev-parse", "HEAD"], { cwd: repository })
+  if (revision.status !== 0) throw new Error("Failed to read the benchmark Git revision")
+  const status = spawnSync("git", ["status", "--porcelain"], { cwd: repository })
+  if (status.status !== 0) throw new Error("Failed to read the benchmark Git status")
+  const bun = spawnSync("bun", ["--version"], { cwd: repository })
+  if (bun.status !== 0) throw new Error("Failed to read the benchmark Bun version")
+  return {
+    arch: process.arch,
+    command: "bun dev:desktop",
+    runs,
+    profile: "fresh",
+    service: "isolated-cold",
+    install: "complete",
+    viteCache: "cold",
+    electronInstall: "present",
+    bunVersion: bun.stdout.toString().trim(),
+    electronVersion: pkg.devDependencies.electron,
+    electronViteVersionRange: pkg.devDependencies["electron-vite"],
+    gitCommit: revision.stdout.toString().trim(),
+    gitDirty: status.stdout.length > 0,
+    trace: Boolean(process.env.OPENCODE_PERFORMANCE_TRACE_DIR),
+  }
+}
+
+export function summarizeDesktopStartup(samples: DesktopStartupSample[]) {
+  return {
+    commandToHomeReadyMs: statistics(samples.map((sample) => sample.commandToHomeReadyMs)),
+    milestonesMs: Object.fromEntries(
+      milestones.map((name) => [name, statistics(samples.map((sample) => sample.milestonesMs[name]))]),
+    ),
+    phasesMs: Object.fromEntries(
+      phases.map((name) => [name, statistics(samples.map((sample) => sample.phasesMs[name]))]),
+    ),
+  }
+}
+
+export function milestoneForLine(line: string): Milestone | undefined {
+  const text = stripAnsi(line)
+  return milestonePatterns.find((item) => text.includes(item.text))?.name
+}
+
+const milestonePatterns: ReadonlyArray<{ name: Milestone; text: string }> = [
+  { name: "bunRootScript", text: "$ bun --cwd packages/desktop dev" },
+  { name: "bunDesktopScript", text: "$ bun ./scripts/dev.ts" },
+  { name: "desktopPrepared", text: "Copied dev icons from" },
+  { name: "mainBundleReady", text: "electron main process built successfully" },
+  { name: "preloadBundleReady", text: "electron preload scripts built successfully" },
+  { name: "rendererDevServerReady", text: "dev server running for the electron renderer process at:" },
+  { name: "electronSpawnStarted", text: "starting electron app..." },
+  { name: "debugEndpointReady", text: "DevTools listening on ws://" },
+  { name: "electronStarted", text: "app starting" },
+  { name: "serviceEnsureStarted", text: "starting v2 background service" },
+  { name: "serviceSpawnRequested", text: "v2 CLI background service starting" },
+  { name: "serviceReady", text: "v2 CLI background service ready" },
+  { name: "backgroundLoadingReady", text: "loading task finished" },
+  { name: "rendererViteConnected", text: "[vite] connected." },
+  { name: "rendererInitializationStarted", text: "awaiting server ready" },
+  { name: "rendererInitializationReady", text: "server ready" },
+  { name: "windowVisible", text: "main window visible" },
+]
+
+async function createColdProfile() {
+  await Promise.all(
+    ["packages/desktop/node_modules/.vite", "packages/desktop/out"].map((path) =>
+      rm(join(repository, path), { recursive: true, force: true }),
+    ),
+  )
+  const root = await mkdtemp(join(tmpdir(), "opencode-desktop-startup-"))
+  return initializeColdProfile(root).catch(async (error) => {
+    await rm(root, { recursive: true, force: true, maxRetries: 5, retryDelay: 100 })
+    throw error
+  })
+}
+
+async function initializeColdProfile(root: string) {
+  await Promise.all(
+    ["data", "config", "cache", "state", "desktop", "session", "home"].map((dir) =>
+      mkdir(join(root, dir), { recursive: true }),
+    ),
+  )
+  await Promise.all([
+    writeFile(
+      join(root, "desktop", "opencode.settings"),
+      JSON.stringify({ firstLaunchOnboardingComplete: true }),
+    ),
+    writeFile(join(root, "desktop", "opencode.global.dat"), JSON.stringify({ language: '{"locale":"en"}' })),
+  ])
+  const registration = join(root, "desktop", "opencode", "service-local.json")
+  await Service.stop({ file: registration })
+  return { root, registration }
+}
+
+function startDesktop(profile: Awaited<ReturnType<typeof createColdProfile>>) {
+  const started = performance.now()
+  const child = spawn("bun", ["dev:desktop"], {
+    cwd: repository,
+    detached: process.platform !== "win32",
+    env: {
+      ...process.env,
+      OPENCODE_CONFIG_DIR: join(profile.root, "config"),
+      OPENCODE_DB: join(profile.root, "data", "opencode.db"),
+      OPENCODE_TEST_HOME: join(profile.root, "home"),
+      OPENCODE_TEST_ONBOARDING: "0",
+      OPENCODE_DESKTOP_TEST_ROOT: profile.root,
+      OPENCODE_DESKTOP_REMOTE_DEBUGGING_PORT: "0",
+      OPENCODE_DESKTOP_DISABLE_PROTOCOL_REGISTRATION: "1",
+    },
+    stdio: ["ignore", "pipe", "pipe"],
+  })
+  if (!child.pid || !child.stdout || !child.stderr) throw new Error("Failed to start the desktop command")
+  const exited = childExit(child)
+  const observed: Partial<Record<Milestone, number>> = {}
+  const endpoint = Promise.withResolvers<string>()
+  const pageErrors: string[] = []
+  let browser: Browser | undefined
+  let service: ServiceInfo | undefined
+  const mark = (name: Milestone) => {
+    observed[name] ??= elapsed(started)
+  }
+  const record = (line: string) => {
+    const milestone = milestoneForLine(line)
+    if (milestone) mark(milestone)
+    const match = stripAnsi(line).match(/DevTools listening on (ws:\/\/\S+)/)
+    if (match?.[1]) endpoint.resolve(match[1])
+  }
+  const stdout = observeOutput(child.stdout, record)
+  const stderr = observeOutput(child.stderr, record)
+
+  return {
+    mark,
+    async open() {
+      const url = await Promise.race([
+        endpoint.promise,
+        exited.then((code) => {
+          throw new Error(`Desktop command exited with code ${code} before opening its debug endpoint`)
+        }),
+        sleep(120_000).then(() => {
+          throw new Error("Timed out waiting for the desktop debug endpoint")
+        }),
+      ])
+      browser = await chromium.connectOverCDP(url, { timeout: 120_000 })
+      const context = browser.contexts()[0]
+      if (!context) throw new Error("Electron did not expose a browser context")
+      await expect.poll(() => context.pages().length, { timeout: 120_000 }).toBeGreaterThan(0)
+      const page = context.pages()[0]
+      if (!page) throw new Error("Electron did not expose a renderer page")
+      page.on("pageerror", (error) => pageErrors.push(error.stack ?? error.message))
+      return page
+    },
+    async result(run: number): Promise<DesktopStartupSample> {
+      if (pageErrors.length) throw new Error(`Desktop renderer reported errors:\n\n${pageErrors.join("\n\n")}`)
+      service = await readService(profile)
+      const milestonesMs = requireMilestones(observed)
+      return {
+        run,
+        commandToHomeReadyMs: milestonesMs.homeReady,
+        milestonesMs,
+        phasesMs: calculatePhases(milestonesMs),
+        service: {
+          version: service.version,
+          url: service.url,
+          pid: service.pid,
+        },
+      }
+    },
+    async close(testInfo: TestInfo, run: number) {
+      const errors: unknown[] = []
+      await browser?.close().catch(() => undefined)
+      await stopProcessTree(child, exited).catch((error) => {
+        errors.push(error)
+        child.stdout?.destroy()
+        child.stderr?.destroy()
+      })
+      const [stdoutText, stderrText] = await Promise.all([stdout, stderr]).catch((error) => {
+        errors.push(error)
+        return ["", ""]
+      })
+      await Promise.all([
+        testInfo.attach(`desktop-startup-${run}-stdout`, { body: stdoutText, contentType: "text/plain" }),
+        testInfo.attach(`desktop-startup-${run}-stderr`, { body: stderrText, contentType: "text/plain" }),
+        pageErrors.length
+          ? testInfo.attach(`desktop-startup-${run}-page-errors`, {
+              body: pageErrors.join("\n\n"),
+              contentType: "text/plain",
+            })
+          : Promise.resolve(),
+      ]).catch((error) => errors.push(error))
+      await Service.stop({ file: profile.registration }).catch((error) => errors.push(error))
+      if (service && processAlive(service.pid))
+        errors.push(new Error(`Desktop service process ${service.pid} did not stop`))
+      await rm(profile.root, { recursive: true, force: true, maxRetries: 5, retryDelay: 100 }).catch((error) =>
+        errors.push(error),
+      )
+      if (errors.length) throw new AggregateError(errors, "Desktop benchmark cleanup failed")
+    },
+  }
+}
+
+async function waitForHome(page: Page, mark: (name: Milestone) => void) {
+  await expect.poll(() => page.evaluate(() => document.visibilityState), { timeout: 120_000 }).toBe("visible")
+
+  const projects = page.getByRole("complementary", { name: "Projects", exact: true })
+  const sessions = page.getByRole("region", { name: "Recent sessions", exact: true })
+  const search = page.getByRole("textbox", { name: "Search sessions", exact: true })
+  const addProject = projects.locator('button[data-action="home-add-project-row"]')
+  await expect(projects).toBeVisible({ timeout: 120_000 })
+  await expect(sessions).toBeVisible()
+  await expect(search).toBeEditable()
+  await expect(sessions.getByText("Nothing here yet", { exact: true })).toBeVisible()
+  await expect(addProject).toHaveCount(1)
+  await addProject.click({ trial: true })
+  mark("homeReady")
+}
+
+type ThemeWindow = Window & {
+  __OPENCODE_THEME_STATES__?: string[]
+  __OPENCODE_THEME_OBSERVER__?: MutationObserver
+}
+
+async function startThemeObservation(page: Page) {
+  await page.addInitScript(installThemeObservation)
+  await page.evaluate(installThemeObservation)
+}
+
+async function requireStableTheme(page: Page) {
+  const states = await page.evaluate(() => {
+    const target = window as ThemeWindow
+    target.__OPENCODE_THEME_OBSERVER__?.disconnect()
+    return target.__OPENCODE_THEME_STATES__ ?? []
+  })
+  if (states.length !== 1) throw new Error(`Desktop theme changed during startup: ${states.join(" -> ")}`)
+}
+
+function installThemeObservation() {
+  const target = window as ThemeWindow
+  const observeRoot = () => {
+    const root = document.documentElement
+    if (!root) return false
+    const state = () => {
+      const theme = root.dataset.theme
+      const scheme = root.dataset.colorScheme
+      return theme && scheme ? `${theme}:${scheme}` : undefined
+    }
+    const initial = state()
+    target.__OPENCODE_THEME_STATES__ = initial ? [initial] : []
+    target.__OPENCODE_THEME_OBSERVER__ = new MutationObserver(() => {
+      const next = state()
+      if (!next) return
+      if (target.__OPENCODE_THEME_STATES__?.at(-1) !== next) target.__OPENCODE_THEME_STATES__?.push(next)
+    })
+    target.__OPENCODE_THEME_OBSERVER__.observe(root, {
+      attributes: true,
+      attributeFilter: ["data-theme", "data-color-scheme"],
+    })
+    return true
+  }
+  if (observeRoot()) return
+  const documentObserver = new MutationObserver(() => {
+    if (!observeRoot()) return
+    documentObserver.disconnect()
+  })
+  target.__OPENCODE_THEME_OBSERVER__ = documentObserver
+  documentObserver.observe(document, { childList: true })
+}
+
+async function observeOutput(stream: NodeJS.ReadableStream, record: (line: string) => void) {
+  const decoder = new TextDecoder()
+  const output: string[] = []
+  let pending = ""
+  for await (const chunk of stream) {
+    const text = typeof chunk === "string" ? chunk : decoder.decode(chunk, { stream: true })
+    output.push(text)
+    pending += text
+    const lines = pending.split(/\r?\n/)
+    pending = lines.pop() ?? ""
+    lines.forEach(record)
+  }
+  const final = decoder.decode()
+  output.push(final)
+  pending += final
+  if (pending) record(pending)
+  return output.join("")
+}
+
+async function readService(profile: Awaited<ReturnType<typeof createColdProfile>>) {
+  const value: unknown = JSON.parse(await readFile(profile.registration, "utf8"))
+  if (!isServiceInfo(value)) throw new Error("Desktop service registration is invalid")
+  const url = new URL(value.url)
+  const port = Number(url.port)
+  if (url.hostname !== "127.0.0.1" || !Number.isInteger(port) || port <= 0)
+    throw new Error(`Desktop service used unexpected endpoint ${value.url}`)
+  if (!value.version.startsWith("2.0.0-local-"))
+    throw new Error(`Desktop service used unexpected version ${value.version}`)
+  return value
+}
+
+function isServiceInfo(value: unknown): value is ServiceInfo {
+  return (
+    typeof value === "object" &&
+    value !== null &&
+    "id" in value &&
+    typeof value.id === "string" &&
+    "version" in value &&
+    typeof value.version === "string" &&
+    "url" in value &&
+    typeof value.url === "string" &&
+    "pid" in value &&
+    typeof value.pid === "number"
+  )
+}
+
+function requireMilestones(observed: Partial<Record<Milestone, number>>) {
+  const get = (name: Milestone) => {
+    const value = observed[name]
+    if (value === undefined) throw new Error(`Desktop startup did not report milestone: ${name}`)
+    return round(value)
+  }
+  return {
+    bunRootScript: get("bunRootScript"),
+    bunDesktopScript: get("bunDesktopScript"),
+    desktopPrepared: get("desktopPrepared"),
+    mainBundleReady: get("mainBundleReady"),
+    preloadBundleReady: get("preloadBundleReady"),
+    rendererDevServerReady: get("rendererDevServerReady"),
+    electronSpawnStarted: get("electronSpawnStarted"),
+    debugEndpointReady: get("debugEndpointReady"),
+    electronStarted: get("electronStarted"),
+    serviceEnsureStarted: get("serviceEnsureStarted"),
+    serviceSpawnRequested: get("serviceSpawnRequested"),
+    serviceReady: get("serviceReady"),
+    backgroundLoadingReady: get("backgroundLoadingReady"),
+    rendererViteConnected: get("rendererViteConnected"),
+    rendererInitializationStarted: get("rendererInitializationStarted"),
+    rendererInitializationReady: get("rendererInitializationReady"),
+    windowVisible: get("windowVisible"),
+    homeReady: get("homeReady"),
+  }
+}
+
+function calculatePhases(value: Record<Milestone, number>): Record<Phase, number> {
+  return {
+    desktopPreparation: value.desktopPrepared,
+    viteMainBundle: round(value.mainBundleReady - value.desktopPrepared),
+    vitePreloadBundle: round(value.preloadBundleReady - value.mainBundleReady),
+    rendererServerStartup: round(value.rendererDevServerReady - value.preloadBundleReady),
+    electronStartup: round(value.electronStarted - value.electronSpawnStarted),
+    serviceSpawnWait: round(value.serviceSpawnRequested - value.serviceEnsureStarted),
+    serviceProcessStartup: round(value.serviceReady - value.serviceSpawnRequested),
+    rendererStartup: round(value.homeReady - value.rendererViteConnected),
+    visibleWindowToHome: round(value.homeReady - value.windowVisible),
+  }
+}
+
+function statistics(values: number[]) {
+  if (!values.length) throw new Error("Cannot summarize an empty benchmark")
+  const sorted = values.toSorted((left, right) => left - right)
+  const median = medianOf(sorted)
+  return {
+    min: round(sorted[0]),
+    median: round(median),
+    max: round(sorted.at(-1)!),
+    medianAbsoluteDeviation: round(medianOf(sorted.map((value) => Math.abs(value - median)).toSorted((a, b) => a - b))),
+  }
+}
+
+function medianOf(sorted: number[]) {
+  const middle = Math.floor(sorted.length / 2)
+  if (sorted.length % 2) return sorted[middle]
+  return (sorted[middle - 1] + sorted[middle]) / 2
+}
+
+async function stopProcessTree(child: ChildProcess, exited: Promise<number | null>) {
+  if (!child.pid) throw new Error("Desktop command has no process ID")
+  if (process.platform !== "win32") return stopProcessGroup(child.pid, exited)
+  if (child.exitCode !== null || (await exitsWithin(child, exited, 2_000))) return
+  const kill = spawn("taskkill.exe", ["/PID", String(child.pid), "/T", "/F"], {
+    stdio: "ignore",
+  })
+  await childExit(kill)
+  if (await exitsWithin(child, exited, 10_000)) return
+  if (!(await exitsWithin(child, exited, 5_000))) throw new Error(`Desktop command process ${child.pid} did not stop`)
+}
+
+async function stopProcessGroup(pid: number, exited: Promise<number | null>) {
+  await Promise.race([exited, sleep(2_000)])
+  if (!processGroupAlive(pid)) return
+  process.kill(-pid, "SIGTERM")
+  if (await processGroupStopsWithin(pid, 10_000)) return
+  process.kill(-pid, "SIGKILL")
+  if (!(await processGroupStopsWithin(pid, 5_000))) throw new Error(`Desktop command process group ${pid} did not stop`)
+}
+
+async function processGroupStopsWithin(pid: number, timeout: number) {
+  const deadline = Date.now() + timeout
+  while (Date.now() < deadline) {
+    if (!processGroupAlive(pid)) return true
+    await sleep(50)
+  }
+  return !processGroupAlive(pid)
+}
+
+function processGroupAlive(pid: number) {
+  try {
+    process.kill(-pid, 0)
+    return true
+  } catch {
+    return false
+  }
+}
+
+async function exitsWithin(child: ChildProcess, exited: Promise<number | null>, timeout: number) {
+  if (child.exitCode !== null) return true
+  const result = await Promise.race([exited.then(() => true), sleep(timeout).then(() => false)])
+  return result
+}
+
+function childExit(child: ChildProcess) {
+  return new Promise<number | null>((resolve, reject) => {
+    child.once("error", reject)
+    child.once("exit", (code) => resolve(code))
+  })
+}
+
+function sleep(milliseconds: number) {
+  return new Promise<void>((resolve) => setTimeout(resolve, milliseconds))
+}
+
+function processAlive(pid: number) {
+  try {
+    process.kill(pid, 0)
+    return true
+  } catch {
+    return false
+  }
+}
+
+function stripAnsi(value: string) {
+  return value.replace(/\x1b\[[0-?]*[ -/]*[@-~]/g, "")
+}
+
+function elapsed(started: number) {
+  return round(performance.now() - started)
+}
+
+function round(value: number) {
+  return Math.round(value * 100) / 100
+}

packages/app/e2e/performance/devex/playwright.config.ts

@@ -0,0 +1,17 @@
+import { defineConfig } from "@playwright/test"
+
+process.env.OPENCODE_PERFORMANCE_RUN_ID ??= `${new Date().toISOString().replace(/[:.]/g, "-")}-${process.pid}`
+
+export default defineConfig({
+  testDir: ".",
+  testMatch: "desktop-startup-benchmark.spec.ts",
+  outputDir: "../../test-results/performance-devex",
+  timeout: 15 * 60_000,
+  expect: {
+    timeout: 120_000,
+  },
+  fullyParallel: false,
+  workers: 1,
+  reporter: [["html", { outputFolder: "../../playwright-report/performance-devex", open: "never" }], ["line"]],
+  projects: [{ name: "desktop" }],
+})

packages/app/e2e/performance/playwright.config.ts

@@ -7,7 +7,7 @@ process.env.OPENCODE_PERFORMANCE_RUN_ID ??= `${new Date().toISOString().replace(
 export default {
   ...config,
   testDir: ".",
-  testIgnore: "unit/**",
+  testIgnore: ["unit/**", "devex/**"],
   outputDir: "../test-results/performance",
   fullyParallel: false,
   workers: 1,

packages/app/e2e/performance/unit/desktop-startup.test.ts

@@ -0,0 +1,88 @@
+import { describe, expect, test } from "bun:test"
+import { inlineThemePreload } from "../../../vite.js"
+import { milestoneForLine, summarizeDesktopStartup, type DesktopStartupSample } from "../devex/desktop-startup"
+
+describe("desktop startup benchmark", () => {
+  test.each(["/oc-theme-preload.js", "./oc-theme-preload.js"])("inlines %s before the renderer runs", (path) => {
+    const html = inlineThemePreload(`<script id="oc-theme-preload-script" src="${path}"></script>`)
+    expect(html).not.toContain(" src=")
+    expect(html).toContain("opencode-color-scheme")
+  })
+
+  test("recognizes startup milestones in colored output", () => {
+    const cases = [
+      ["bunRootScript", "$ bun --cwd packages/desktop dev"],
+      ["bunDesktopScript", "$ bun ./scripts/dev.ts"],
+      ["desktopPrepared", "Copied dev icons from"],
+      ["mainBundleReady", "electron main process built successfully"],
+      ["preloadBundleReady", "electron preload scripts built successfully"],
+      ["rendererDevServerReady", "dev server running for the electron renderer process at:"],
+      ["electronSpawnStarted", "starting electron app..."],
+      ["debugEndpointReady", "DevTools listening on ws://"],
+      ["electronStarted", "app starting"],
+      ["serviceEnsureStarted", "starting v2 background service"],
+      ["serviceSpawnRequested", "v2 CLI background service starting"],
+      ["serviceReady", "v2 CLI background service ready"],
+      ["backgroundLoadingReady", "loading task finished"],
+      ["rendererViteConnected", "[vite] connected."],
+      ["rendererInitializationStarted", "awaiting server ready"],
+      ["rendererInitializationReady", "server ready"],
+      ["windowVisible", "main window visible"],
+    ] as const
+    cases.forEach(([milestone, line]) => {
+      expect(milestoneForLine(`\u001b[32m${line}\u001b[39m`)).toBe(milestone)
+    })
+    expect(milestoneForLine("12:30:00.000 › v2 CLI background service ready {")).toBe("serviceReady")
+    expect(milestoneForLine("unrelated output")).toBeUndefined()
+  })
+
+  test("keeps raw samples and reports median absolute deviation", () => {
+    const samples = [24, 20, 22, 28, 26].map((commandToHomeReadyMs, index) => sample(index + 1, commandToHomeReadyMs))
+    expect(summarizeDesktopStartup(samples).commandToHomeReadyMs).toEqual({
+      min: 20,
+      median: 24,
+      max: 28,
+      medianAbsoluteDeviation: 2,
+    })
+  })
+})
+
+function sample(run: number, commandToHomeReadyMs: number): DesktopStartupSample {
+  const milestonesMs = {
+    bunRootScript: 1,
+    bunDesktopScript: 2,
+    desktopPrepared: 3,
+    mainBundleReady: 4,
+    preloadBundleReady: 5,
+    rendererDevServerReady: 6,
+    electronSpawnStarted: 7,
+    debugEndpointReady: 8,
+    electronStarted: 9,
+    serviceEnsureStarted: 10,
+    serviceSpawnRequested: 11,
+    serviceReady: 12,
+    backgroundLoadingReady: 13,
+    rendererViteConnected: 14,
+    rendererInitializationStarted: 15,
+    rendererInitializationReady: 16,
+    windowVisible: 17,
+    homeReady: commandToHomeReadyMs,
+  }
+  return {
+    run,
+    commandToHomeReadyMs,
+    milestonesMs,
+    phasesMs: {
+      desktopPreparation: 3,
+      viteMainBundle: 1,
+      vitePreloadBundle: 1,
+      rendererServerStartup: 1,
+      electronStartup: 2,
+      serviceSpawnWait: 1,
+      serviceProcessStartup: 1,
+      rendererStartup: commandToHomeReadyMs - 14,
+      visibleWindowToHome: commandToHomeReadyMs - 17,
+    },
+    service: { version: "2.0.0-local-test", url: "http://127.0.0.1:3000", pid: run },
+  }
+}

packages/app/e2e/regression/prompt-input-v2-command-draft.spec.ts

@@ -36,6 +36,9 @@ test("preserves the draft when a populated command menu triggers a built-in", as
   await page.goto(`/server/${base64Encode(server)}/session/${sessionID}`)
   const composer = page.locator('[data-component="prompt-input-v2"]')
   const input = composer.locator('[data-component="prompt-input"]')
+  await expect.poll(() => input.evaluate((element) => getComputedStyle(element, "::before").content)).toBe(
+    `"${String.fromCodePoint(0x200b)}"`,
+  )
   await expectAppVisible(composer)
 
   await input.fill("keep me")

packages/app/e2e/regression/remote-session-settings.spec.ts

@@ -19,7 +19,7 @@ test("session settings use the remote server context", async ({ page }) => {
   await configureServers(page)
 
   await page.goto(`/server/${base64Encode(serverB)}/session/${sessionB.id}`)
-  await expect(page.getByText(sessionB.title).first()).toBeVisible()
+  await expect(page.getByRole("heading", { name: sessionB.title, exact: true })).toBeVisible()
   await page.keyboard.press("Control+,")
 
   const dialog = page.locator(".settings-v2-dialog")
@@ -61,7 +61,7 @@ test("auto-accept responds for an unfocused server session", async ({ page }) =>
 
   const hrefB = `/server/${base64Encode(serverB)}/session/${sessionB.id}`
   await page.goto(`/server/${base64Encode(serverA)}/session/${sessionA.id}`)
-  await expect(page.getByText(sessionA.title).first()).toBeVisible()
+  await expect(page.getByRole("heading", { name: sessionA.title, exact: true })).toBeVisible()
   await page.keyboard.press("Control+,")
   const autoAccept = page.locator(".settings-v2-dialog").locator('[data-action="settings-auto-accept-permissions"]')
   await autoAccept.locator('[data-slot="switch-control"]').click()
@@ -78,7 +78,7 @@ test("auto-accept responds for an unfocused server session", async ({ page }) =>
 
   await page.locator(`[data-titlebar-tab-slot]:has(a[href="${hrefB}"])`).click()
   await expect(page).toHaveURL(new RegExp(`${hrefB.replace(/[.*+?^${}()|[\]\\]/g, "\\$&")}$`))
-  await expect(page.getByText(sessionB.title).first()).toBeVisible()
+  await expect(page.getByRole("heading", { name: sessionB.title, exact: true })).toBeVisible()
   await transport.waitForConnection()
 
   await transport.send({
@@ -208,7 +208,7 @@ async function mockServers(page: Page, permissionRequests: string[], permissionR
       return json(route, [
         {
           id: remote ? sessionB.projectID : "project-server-a",
-          worktree: directory,
+          canonical: directory,
           vcs: "git",
           time: { created: 1, updated: 1 },
           sandboxes: [],
@@ -216,7 +216,7 @@ async function mockServers(page: Page, permissionRequests: string[], permissionR
       ])
     }
     if (url.pathname === "/api/project/current")
-      return json(route, { id: remote ? sessionB.projectID : "project-server-a", directory })
+      return json(route, { id: remote ? sessionB.projectID : "project-server-a", directory, canonical: directory })
     if (url.pathname === "/api/session")
       return json(route, { data: sessions.map((session) => currentSession(session)), cursor: {} })
     if (url.pathname === "/api/session/active") return json(route, { data: {} })
@@ -237,7 +237,7 @@ function session(id: string, directory: string, title: string) {
     id,
     slug: id,
     projectID: `project-${id}`,
-    directory,
+    location: { directory },
     title,
     version: "dev",
     time: { created: 1, updated: 1 },

packages/app/package.json

@@ -5,6 +5,7 @@
   "type": "module",
   "exports": {
     ".": "./src/index.ts",
+    "./desktop": "./src/desktop.ts",
     "./desktop-menu": "./src/desktop-menu.ts",
     "./i18n/desktop-native": "./src/i18n/desktop-native.ts",
     "./updater": "./src/updater.ts",
@@ -28,14 +29,15 @@
     "test:e2e:ui": "playwright test --ui",
     "test:e2e:report": "playwright show-report e2e/playwright-report",
     "test:stability": "bun test ./e2e/performance/unit/visual-stability.test.ts && playwright test --config e2e/performance/timeline-stability/playwright.config.ts",
-    "test:bench": "bun test ./e2e/performance/unit && playwright test --config e2e/performance/playwright.config.ts"
+    "test:bench": "bun test ./e2e/performance/unit && playwright test --config e2e/performance/playwright.config.ts",
+    "test:bench:devex": "bun test ./e2e/performance/unit/desktop-startup.test.ts && playwright test --config e2e/performance/devex/playwright.config.ts"
   },
   "license": "MIT",
   "devDependencies": {
     "@happy-dom/global-registrator": "20.0.11",
     "@playwright/test": "catalog:",
     "@sentry/vite-plugin": "catalog:",
-    "@tailwindcss/vite": "catalog:",
+    "@tailwindcss/vite": "4.3.3",
     "@tsconfig/bun": "1.0.9",
     "@types/bun": "catalog:",
     "@types/luxon": "catalog:",
@@ -44,9 +46,9 @@
     "happy-dom": "20.11.1",
     "tw-animate-css": "1.4.0",
     "typescript": "catalog:",
-    "vite": "catalog:",
+    "vite": "8.2.1",
     "vite-plugin-icons-spritesheet": "3.0.1",
-    "vite-plugin-solid": "catalog:"
+    "vite-plugin-solid": "2.11.14"
   },
   "dependencies": {
     "@corvu/drawer": "catalog:",
@@ -85,6 +87,6 @@
     "solid-js": "catalog:",
     "solid-list": "catalog:",
     "solid-presence": "0.2.0",
-    "tailwindcss": "catalog:"
+    "tailwindcss": "4.3.3"
   }
 }

packages/app/src/app.tsx

@@ -1,12 +1,10 @@
 import "@/index.css"
-import * as Sentry from "@sentry/solid"
 import { DialogProvider } from "@opencode-ai/ui/context/dialog"
 import { FileComponentProvider } from "@opencode-ai/ui/context/file"
-import { File } from "@opencode-ai/session-ui/file"
 import { Font } from "@opencode-ai/ui/font"
 import { ThemeProvider } from "@opencode-ai/ui/theme/context"
 import { MetaProvider } from "@solidjs/meta"
-import { type BaseRouterProps, Navigate, Route, Router, useParams, useSearchParams } from "@solidjs/router"
+import { type BaseRouterProps, Route, Router, useParams } from "@solidjs/router"
 import { QueryClient, QueryClientProvider } from "@tanstack/solid-query"
 import {
   type Component,
@@ -20,30 +18,37 @@ import {
 } from "solid-js"
 import { Dynamic } from "solid-js/web"
 import { CommandProvider, useCommand, type CommandOption } from "@/context/command"
-import { CommentsProvider } from "@/context/comments"
-import { FileProvider } from "@/context/file"
 import { GlobalProvider, useGlobal } from "@/context/global"
 import { HighlightsProvider } from "@/context/highlights"
 import { LanguageProvider, UiI18nBridge, type Locale, useLanguage } from "@/context/language"
 import { LayoutProvider } from "@/context/layout"
-import { ModelsProvider } from "@/context/models"
 import { usePlatform } from "@/context/platform"
-import { PromptProvider } from "@/context/prompt"
 import { ServerConnection, ServersProvider } from "@/context/servers"
 import { SettingsProvider } from "@/context/settings"
-import { TabsProvider, useTabs, type DraftTab } from "@/context/tabs"
-import { LocationProvider } from "@/context/location"
+import { TabsProvider } from "@/context/tabs"
 import { WslServersProvider } from "@/wsl/context"
-import { SessionUIProvider } from "@/pages/directory-layout"
 import Layout from "@/pages/layout"
 import { ErrorPage } from "./pages/error"
 import { requireServerKey } from "./utils/session-route"
 
-import { TargetSessionRouteContent } from "@/pages/session"
 import { Home } from "@/pages/home"
 import { ServerProvider } from "./context/server"
 
-const NewSession = lazy(() => import("@/pages/new-session"))
+const File = lazy(() => import("@opencode-ai/session-ui/file").then((module) => ({ default: module.File })))
+const loadDraftRoute = () => Promise.all([import("@/pages/draft-route"), File.preload()]).then(([module]) => module)
+const loadSessionRoute = () => Promise.all([import("@/pages/session"), File.preload()]).then(([module]) => module)
+const DraftRoute = lazy(() => loadDraftRoute().then((module) => ({ default: module.DraftRoute })))
+const TargetSessionRouteContent = lazy(() =>
+  loadSessionRoute().then((module) => ({ default: module.TargetSessionRouteContent })),
+)
+
+export function preloadRoute(url: string) {
+  const pathname = url.split(/[?#]/, 1)[0]
+  if (pathname === "/new-session") return DraftRoute.preload().then(() => undefined)
+  if (/^\/server\/[^/]+\/session\/[^/]+$/.test(pathname))
+    return TargetSessionRouteContent.preload().then(() => undefined)
+  return Promise.resolve()
+}
 
 function TargetServerRoute(props: ParentProps) {
   const params = useParams<{ serverKey: string }>()
@@ -60,45 +65,6 @@ function TargetServerRoute(props: ParentProps) {
   )
 }
 
-function DraftRoute() {
-  const [search] = useSearchParams<{ draftId?: string }>()
-  const tabs = useTabs()
-  return (
-    <Show
-      when={tabs.store.find((tab): tab is DraftTab => tab.type === "draft" && tab.draftID === search.draftId)}
-      keyed
-      fallback={tabs.ready() && <Navigate href="/" />}
-    >
-      {(draft) => <ResolvedDraftRoute draft={draft} />}
-    </Show>
-  )
-}
-
-function ResolvedDraftRoute(props: { draft: DraftTab }) {
-  const global = useGlobal()
-  const conn = createMemo(() => global.servers.list().find((item) => ServerConnection.key(item) === props.draft.server))
-
-  return (
-    <Show when={`${props.draft.server}\0${props.draft.directory}`} keyed>
-      <Show when={conn()} keyed>
-        {(conn) => (
-          <ServerProvider conn={conn}>
-            <ModelsProvider directory={props.draft.directory}>
-              <LocationProvider directory={props.draft.directory}>
-                <SessionUIProvider directory={props.draft.directory} server={props.draft.server}>
-                  <DraftProviders>
-                    <NewSession draftId={props.draft.draftID} />
-                  </DraftProviders>
-                </SessionUIProvider>
-              </LocationProvider>
-            </ModelsProvider>
-          </ServerProvider>
-        )}
-      </Show>
-    </Show>
-  )
-}
-
 declare global {
   interface Window {
     __OPENCODE__?: {
@@ -167,22 +133,11 @@ function AppLayout(props: ParentProps) {
   )
 }
 
-// The draft page only renders the prompt composer, so it drops TerminalProvider.
-// FileProvider and CommentsProvider stay because PromptInput uses file search and comment context.
-function DraftProviders(props: ParentProps) {
-  return (
-    <FileProvider>
-      <PromptProvider>
-        <CommentsProvider>{props.children}</CommentsProvider>
-      </PromptProvider>
-    </FileProvider>
-  )
-}
-
 export function AppBaseProviders(
   props: ParentProps<{
     locale?: Locale
     onNativeTranslations?: Parameters<typeof LanguageProvider>[0]["onNativeTranslations"]
+    onThemeApplied?: () => void
   }>,
 ) {
   return (
@@ -191,13 +146,14 @@ export function AppBaseProviders(
       <ThemeProvider
         onThemeApplied={(_, mode, scheme) => {
           void window.api?.setTitlebar?.({ mode, scheme })
+          props.onThemeApplied?.()
         }}
       >
         <LanguageProvider locale={props.locale} onNativeTranslations={props.onNativeTranslations}>
           <UiI18nBridge>
             <ErrorBoundary
               fallback={(error) => {
-                Sentry.captureException(error)
+                void import("@sentry/solid").then(({ captureException }) => captureException(error))
                 return <ErrorPage error={error} />
               }}
             >

packages/app/src/components/titlebar.tsx

@@ -241,7 +241,11 @@ export function Titlebar(props: { update?: TitlebarUpdate; debugTools?: { visibl
                     sessionId: activeSession.id,
                   }
                   const model = tabs.stateValue<PromptSession>(sessionTab, "prompt")?.model.current()
-                  tabs.newDraft({ server: sessionTab.server, directory: activeSession.location.directory }, "", model)
+                  void tabs.newDraft(
+                    { server: sessionTab.server, directory: activeSession.location.directory },
+                    "",
+                    model,
+                  )
                   return
                 }
                 case "draft": {
@@ -249,7 +253,7 @@ export function Titlebar(props: { update?: TitlebarUpdate; debugTools?: { visibl
                   if (activeTab?.type !== "draft") return
 
                   const model = tabs.stateValue<PromptSession>(activeTab, "prompt")?.model.current()
-                  tabs.newDraft({ server: activeTab.server, directory: activeTab.directory }, "", model)
+                  void tabs.newDraft({ server: activeTab.server, directory: activeTab.directory }, "", model)
                   return
                 }
                 case "home": {
@@ -263,7 +267,7 @@ export function Titlebar(props: { update?: TitlebarUpdate; debugTools?: { visibl
                     projects?.list().find((item) => item.worktree === projects.last()) ??
                     projects?.list()[0]
                   if (conn && project) {
-                    tabs.newDraft({ server: ServerConnection.key(conn), directory: project.worktree }, "")
+                    void tabs.newDraft({ server: ServerConnection.key(conn), directory: project.worktree }, "")
                     return
                   }
                 }
@@ -467,11 +471,14 @@ function ChannelIndicator(props: { debugTools?: { visible: boolean; toggle: () =
     )
   }
 
+  const label = channel && ["local", "beta", "dev"].includes(channel) ? channel.toUpperCase() : undefined
   return (
-    <Show when={["local", "beta", "dev"].includes(channel)}>
-      <div class="bg-icon-interactive-base text-[#FFF] font-medium px-2 rounded-sm uppercase font-mono">
-        {channel.toUpperCase()}
-      </div>
+    <Show when={label}>
+      {(value) => (
+        <div class="bg-icon-interactive-base text-[#FFF] font-medium px-2 rounded-sm uppercase font-mono">
+          {value()}
+        </div>
+      )}
     </Show>
   )
 }

packages/app/src/desktop.ts

@@ -0,0 +1,9 @@
+export { AppBaseProviders, AppInterface, preloadRoute } from "./app"
+export { ACCEPTED_FILE_EXTENSIONS } from "./constants/file-picker"
+export { useCommand } from "./context/command"
+export { loadLocaleDict, normalizeLocale, type Locale, useLanguage } from "./context/language"
+export { type Platform, PlatformProvider } from "./context/platform"
+export { ServerConnection, useServers } from "./context/servers"
+export { useTabs } from "./context/tabs"
+export { createDraftStore } from "./utils/draft-store"
+export { useWslServers } from "./wsl/context"

packages/app/src/index.ts

@@ -1,4 +1,4 @@
-export { AppBaseProviders, AppInterface } from "./app"
+export { AppBaseProviders, AppInterface, preloadRoute } from "./app"
 export { useLayout } from "./context/layout"
 export { useServerSDK } from "./context/server-sdk"
 export { useServers as useServers } from "./context/servers"

packages/app/src/pages/draft-route.tsx

@@ -0,0 +1,64 @@
+import { Navigate, useSearchParams } from "@solidjs/router"
+import { createMemo, Show, type ParentProps } from "solid-js"
+import { CommentsProvider } from "@/context/comments"
+import { FileProvider } from "@/context/file"
+import { useGlobal } from "@/context/global"
+import { LocationProvider } from "@/context/location"
+import { ModelsProvider } from "@/context/models"
+import { PromptProvider } from "@/context/prompt"
+import { ServerProvider } from "@/context/server"
+import { ServerConnection } from "@/context/servers"
+import { useTabs, type DraftTab } from "@/context/tabs"
+import { SessionUIProvider } from "@/pages/directory-layout"
+import NewSession from "@/pages/new-session"
+
+export function DraftRoute() {
+  const [search] = useSearchParams<{ draftId?: string }>()
+  const tabs = useTabs()
+  return (
+    <Show
+      when={tabs.store.find((tab): tab is DraftTab => tab.type === "draft" && tab.draftID === search.draftId)}
+      keyed
+      fallback={tabs.ready() && <Navigate href="/" />}
+    >
+      {(draft) => <ResolvedDraftRoute draft={draft} />}
+    </Show>
+  )
+}
+
+function ResolvedDraftRoute(props: { draft: DraftTab }) {
+  const global = useGlobal()
+  const conn = createMemo(() => global.servers.list().find((item) => ServerConnection.key(item) === props.draft.server))
+
+  return (
+    <Show when={`${props.draft.server}\0${props.draft.directory}`} keyed>
+      <Show when={conn()} keyed>
+        {(conn) => (
+          <ServerProvider conn={conn}>
+            <ModelsProvider directory={props.draft.directory}>
+              <LocationProvider directory={props.draft.directory}>
+                <SessionUIProvider directory={props.draft.directory} server={props.draft.server}>
+                  <DraftProviders>
+                    <NewSession draftId={props.draft.draftID} />
+                  </DraftProviders>
+                </SessionUIProvider>
+              </LocationProvider>
+            </ModelsProvider>
+          </ServerProvider>
+        )}
+      </Show>
+    </Show>
+  )
+}
+
+// The draft page only renders the prompt composer, so it drops TerminalProvider.
+// FileProvider and CommentsProvider stay because PromptInput uses file search and comment context.
+function DraftProviders(props: ParentProps) {
+  return (
+    <FileProvider>
+      <PromptProvider>
+        <CommentsProvider>{props.children}</CommentsProvider>
+      </PromptProvider>
+    </FileProvider>
+  )
+}

packages/app/src/pages/home/home-projects-controller.tsx

@@ -1,10 +1,8 @@
 import { useDirectoryPicker } from "@/components/directory-picker"
 import { useServerActionsController } from "@/components/server/server-management-controller"
 import { useSettingsCommand } from "@/components/settings-dialog"
-import { DialogServerV2 } from "@/components/settings-v2/dialog-server-v2"
 import { type LocalProject } from "@/context/layout"
 import { useLanguage } from "@/context/language"
-import { useNotification } from "@/context/notification"
 import { usePlatform } from "@/context/platform"
 import { ServerConnection } from "@/context/servers"
 import { closeHomeProject, errorMessage, homeProjectDirectories } from "@/pages/layout/helpers"
@@ -63,7 +61,11 @@ export function createHomeProjectsController(home: HomeController) {
         serverManagement.defaults.set(conn ? ServerConnection.key(conn) : null),
       canRemove: (conn: ServerConnection.Any) => serverManagement.connection.canRemove(ServerConnection.key(conn)),
       remove: (conn: ServerConnection.Any) => serverManagement.connection.remove(ServerConnection.key(conn)),
-      edit: (conn: ServerConnection.Http) => dialog.show(() => <DialogServerV2 mode="edit" server={conn} />),
+      edit: (conn: ServerConnection.Http) => {
+        void import("@/components/settings-v2/dialog-server-v2").then(({ DialogServerV2 }) => {
+          void dialog.show(() => <DialogServerV2 mode="edit" server={conn} />)
+        })
+      },
       focus: home.selection.focusServer,
     },
     project: {

packages/app/src/pages/layout.tsx

@@ -1,13 +1,14 @@
-import { Show, Suspense, type ParentProps } from "solid-js"
+import { lazy, Show, Suspense, type ParentProps } from "solid-js"
 import { createStore } from "solid-js/store"
-import { DebugBar } from "@/components/debug-bar"
 import { Titlebar, type TitlebarUpdate } from "@/components/titlebar"
 import { usePlatform } from "@/context/platform"
 import { ToastRegion } from "@/utils/toast"
 
+const DebugBar = lazy(() => import("@/components/debug-bar").then((module) => ({ default: module.DebugBar })))
+
 export default function Layout(props: ParentProps) {
   const platform = usePlatform()
-  const [state, setState] = createStore({ debugTools: true })
+  const [state, setState] = createStore({ debugTools: false })
 
   const update: TitlebarUpdate = {
     get version() {
@@ -41,7 +42,9 @@ export default function Layout(props: ParentProps) {
         <Suspense>{props.children}</Suspense>
       </main>
       <Show when={import.meta.env.DEV && state.debugTools}>
-        <DebugBar inline />
+        <Suspense>
+          <DebugBar inline />
+        </Suspense>
       </Show>
       <ToastRegion />
     </div>

packages/app/src/pages/new-session/new-session-workspace-controller.test.ts

@@ -39,14 +39,20 @@ describe("new session workspace selection", () => {
     expect(normalizeNewSessionWorktree("main", "C:\\Repo\\", "c:/repo")).toBe("main")
   })
 
-  test("falls back to the local branch for main, create, and unknown worktrees", () => {
+  test("resolves the branch from the active location", () => {
     const branch = (worktree: string) => (worktree === "/project/feature" ? "feature" : undefined)
-    expect(resolveNewSessionBranch({ worktree: "main", local: "dev", worktreeBranch: branch })).toBe("dev")
-    expect(resolveNewSessionBranch({ worktree: "create", local: "dev", worktreeBranch: branch })).toBe("dev")
-    expect(resolveNewSessionBranch({ worktree: "/project/feature", local: "dev", worktreeBranch: branch })).toBe(
+    expect(resolveNewSessionBranch({ worktree: "main", directory: "/project/feature", worktreeBranch: branch })).toBe(
       "feature",
     )
-    expect(resolveNewSessionBranch({ worktree: "/missing", local: "dev", worktreeBranch: branch })).toBe("dev")
+    expect(
+      resolveNewSessionBranch({ worktree: "create", directory: "/project/feature", worktreeBranch: branch }),
+    ).toBe("feature")
+    expect(
+      resolveNewSessionBranch({ worktree: "/project/feature", directory: "/project", worktreeBranch: branch }),
+    ).toBe("feature")
+    expect(resolveNewSessionBranch({ worktree: "/missing", directory: "/project/feature", worktreeBranch: branch })).toBe(
+      undefined,
+    )
   })
 
   test("uses location VCS state when the project inventory is stale", () => {

packages/app/src/pages/new-session/new-session-workspace-controller.ts

@@ -32,11 +32,11 @@ export function normalizeNewSessionWorktree(value: string, directory: string, pr
 
 export function resolveNewSessionBranch(input: {
   worktree: string
-  local?: string
+  directory: string
   worktreeBranch: (worktree: string) => string | undefined
 }) {
-  if (input.worktree === "main" || input.worktree === "create") return input.local
-  return input.worktreeBranch(input.worktree) ?? input.local
+  const directory = input.worktree === "main" || input.worktree === "create" ? input.directory : input.worktree
+  return input.worktreeBranch(directory)
 }
 
 export function resolveNewSessionGit(input: { projectVcs?: string; branch?: string }) {
@@ -95,11 +95,10 @@ export function createNewSessionWorkspaceController(input: {
     const directories = project ? [project.worktree, ...workspaceDirectories(project)] : [sdk().directory]
     directories.forEach((directory) => void data.location.vcs.sync({ directory }).catch(() => undefined))
   })
-  const localBranch = createMemo(() => data.location.vcs.info({ directory: projectRoot() })?.branch.current)
   const branch = createMemo(() =>
     resolveNewSessionBranch({
       worktree: value(),
-      local: localBranch(),
+      directory: sdk().directory,
       worktreeBranch: (worktree) => data.location.vcs.info({ directory: worktree })?.branch.current,
     }),
   )

packages/app/src/theme-preload.test.ts

@@ -3,22 +3,56 @@ import { beforeEach, describe, expect, test } from "bun:test"
 const src = await Bun.file(new URL("../public/oc-theme-preload.js", import.meta.url)).text()
 
 const run = () => Function(src)()
+const setSystemDark = (matches: boolean) =>
+  Object.defineProperty(window, "matchMedia", {
+    value: () => ({ matches }) as MediaQueryList,
+    configurable: true,
+  })
 
 beforeEach(() => {
   document.head.innerHTML = ""
   document.documentElement.removeAttribute("data-theme")
   document.documentElement.removeAttribute("data-color-scheme")
+  document.documentElement.style.removeProperty("background-color")
   localStorage.clear()
-  Object.defineProperty(window, "matchMedia", {
-    value: () =>
-      ({
-        matches: false,
-      }) as MediaQueryList,
-    configurable: true,
-  })
+  setSystemDark(false)
 })
 
 describe("theme preload", () => {
+  test("uses default theme and system light mode when settings are absent", () => {
+    run()
+
+    expect(document.documentElement.dataset.theme).toBe("oc-2")
+    expect(document.documentElement.dataset.colorScheme).toBe("light")
+    expect(document.documentElement.style.backgroundColor).toBe("#fafafa")
+  })
+
+  test("restores explicit dark mode on a light system", () => {
+    localStorage.setItem("opencode-color-scheme", "dark")
+    run()
+
+    expect(document.documentElement.dataset.colorScheme).toBe("dark")
+    expect(document.documentElement.style.backgroundColor).toBe("#080808")
+  })
+
+  test("restores explicit light mode on a dark system", () => {
+    setSystemDark(true)
+    localStorage.setItem("opencode-color-scheme", "light")
+    run()
+
+    expect(document.documentElement.dataset.colorScheme).toBe("light")
+    expect(document.documentElement.style.backgroundColor).toBe("#fafafa")
+  })
+
+  test("resolves persisted system mode before paint", () => {
+    setSystemDark(true)
+    localStorage.setItem("opencode-color-scheme", "system")
+    run()
+
+    expect(document.documentElement.dataset.colorScheme).toBe("dark")
+    expect(document.documentElement.style.backgroundColor).toBe("#080808")
+  })
+
   test("keeps cached css for non-default themes", () => {
     localStorage.setItem("opencode-theme-id", "nightowl")
     localStorage.setItem("opencode-theme-css-light", "--background-base:#fff;")
@@ -28,4 +62,15 @@ describe("theme preload", () => {
     expect(document.documentElement.dataset.theme).toBe("nightowl")
     expect(document.getElementById("oc-theme-preload")?.textContent).toContain("--background-base:#fff;")
   })
+
+  test("restores the cached variant for a persisted custom dark theme", () => {
+    localStorage.setItem("opencode-theme-id", "nightowl")
+    localStorage.setItem("opencode-color-scheme", "dark")
+    localStorage.setItem("opencode-theme-css-dark", "--background-base:#010203;")
+    run()
+
+    expect(document.documentElement.dataset.theme).toBe("nightowl")
+    expect(document.documentElement.dataset.colorScheme).toBe("dark")
+    expect(document.getElementById("oc-theme-preload")?.textContent).toContain("--background-base:#010203;")
+  })
 })

packages/app/vite.config.ts

@@ -1,6 +1,6 @@
 import { sentryVitePlugin } from "@sentry/vite-plugin"
 import { defineConfig } from "vite"
-import desktopPlugin from "./vite"
+import desktopPlugin from "./vite.js"
 
 const sentry =
   process.env.SENTRY_AUTH_TOKEN && process.env.SENTRY_ORG && process.env.SENTRY_PROJECT

packages/app/vite.js

@@ -4,6 +4,18 @@ import tailwindcss from "@tailwindcss/vite"
 import { fileURLToPath } from "url"
 
 const theme = fileURLToPath(new URL("./public/oc-theme-preload.js", import.meta.url))
+const themeScript = readFileSync(theme, "utf8")
+const tailwind = tailwindcss()
+const tailwindGenerate = tailwind.find((plugin) => plugin.name === "@tailwindcss/vite:generate:serve")
+const tailwindHotUpdate = tailwindGenerate?.hotUpdate
+
+// Tailwind 4.3.3 expects a server that Vite's bundled dev hook does not provide.
+if (tailwindGenerate && typeof tailwindHotUpdate === "function") {
+  tailwindGenerate.hotUpdate = function (context) {
+    if (!context.server) return
+    return tailwindHotUpdate.call(this, context)
+  }
+}
 
 const channel = (() => {
   const raw = process.env.OPENCODE_CHANNEL
@@ -40,13 +52,18 @@ export default [
   },
   {
     name: "opencode-desktop:theme-preload",
-    transformIndexHtml(html) {
-      return html.replace(
-        '<script id="oc-theme-preload-script" src="/oc-theme-preload.js"></script>',
-        `<script id="oc-theme-preload-script">${readFileSync(theme, "utf8")}</script>`,
-      )
+    transformIndexHtml: {
+      order: "pre",
+      handler: inlineThemePreload,
     },
   },
-  tailwindcss(),
+  ...tailwind,
   solidPlugin(),
 ]
+
+export function inlineThemePreload(html) {
+  return html.replace(
+    /<script id="oc-theme-preload-script" src="(?:\.\/|\/)oc-theme-preload\.js"><\/script>/,
+    `<script id="oc-theme-preload-script">${themeScript}</script>`,
+  )
+}

packages/client/package.json

@@ -37,7 +37,7 @@
     "@opencode-ai/protocol": "workspace:*"
   },
   "peerDependencies": {
-    "effect": "4.0.0-beta.107",
+    "effect": "4.0.0-rc.110",
     "solid-js": ">=1.9.0"
   },
   "peerDependenciesMeta": {

packages/client/src/service-timing.ts

@@ -12,8 +12,8 @@ export type EnsureTiming = {
 const timings = new WeakMap<object, EnsureTiming>()
 
 export const defaultEnsureTiming: EnsureTiming = {
-  pollInterval: 1_000,
-  attempts: 120,
+  pollInterval: 100,
+  attempts: 1_200,
   requestTimeout: 2_000,
   spawnDelay: 5_000,
   maxSpawnDelay: 30_000,

packages/client/test/effect.test.ts

@@ -111,7 +111,7 @@ test("event.subscribe exposes and decodes the native Effect event stream", async
   expect(Array.from(events).map((event) => event.type)).toEqual(["server.connected", "session.model.selected"])
   const durable = events[1]
   if (durable?.type !== "session.model.selected") throw new Error("Expected model event")
-  expect(DateTime.toEpochMillis(durable.created)).toBe(1_717_171_717_000)
+  expect(durable.created).toBe(1_717_171_717_000)
   expect(durable.durable).toEqual({ aggregateID: "ses_test", seq: 1, version: 1 })
 })
 
@@ -219,9 +219,7 @@ test("session methods retain decoded Effect inputs and outputs", async () => {
   expect(logQueries[0]).toEqual({ after: "0" })
   const logged = Array.from(result.log)
   expect(logged.map((item) => item.type)).toEqual(["session.model.selected", "log.synced"])
-  expect(logged[0]?.type === "session.model.selected" && DateTime.toEpochMillis(logged[0].created)).toBe(
-    1_717_171_717_000,
-  )
+  expect(logged[0]?.type === "session.model.selected" && logged[0].created).toBe(1_717_171_717_000)
   expect(logged.at(-1)).toEqual(synced)
   expect(result.message).toEqual(expect.objectContaining({ id: "msg_model", type: "model-switched" }))
 })

packages/client/test/fixture/service-timing.ts

@@ -2,6 +2,7 @@ import { withEnsureTiming } from "../../src/service-timing"
 
 const timing = {
   pollInterval: 20,
+  attempts: 120,
   requestTimeout: 100,
   spawnDelay: 200,
   maxSpawnDelay: 1_200,

packages/core/src/aisdk.ts

@@ -328,7 +328,6 @@ function modelFromLanguage(info: Info, language: LanguageModelV3) {
               body: projected.body === undefined ? undefined : { ...projected.body },
               headers: info.headers,
             },
-      limits: { context: info.limit.context, input: info.limit.input, output: info.limit.output },
       providerOptions: projected.settings,
     },
     body: {

packages/core/src/mcp/index.ts

@@ -5,7 +5,7 @@ import { McpEvent } from "@opencode-ai/schema/mcp-event"
 import { Command } from "@opencode-ai/schema/command"
 import { createHash } from "node:crypto"
 import { isDeepStrictEqual } from "node:util"
-import { Cause, Context, Deferred, Effect, Exit, FiberSet, Layer, Schema, Scope, Stream, Types } from "effect"
+import { Cause, Context, Effect, Exit, FiberSet, Latch, Layer, Schema, Scope, Stream, Types } from "effect"
 import { makeLocationNode } from "@opencode-ai/util/effect/app-node"
 import { Credential } from "../credential.js"
 import { Bus } from "../bus.js"
@@ -111,7 +111,7 @@ export class ToolCallError extends Schema.TaggedError<ToolCallError>()("MCP.Tool
 type ServerEntry = {
   readonly config: Mcp.ServerConfig
   status: Status
-  readonly startup: Deferred.Deferred<void>
+  readonly startup: Latch.Latch
   scope?: Scope.Closeable
   client?: MCPClient.Connection
   tools?: ReadonlyArray<Tool>
@@ -535,7 +535,7 @@ export const layer = (options?: Options) =>
               : { status: "failed", error: error instanceof Error ? error.message : String(error) }
           yield* Effect.logWarning("mcp connect failed", { server: name, status: entry.status })
           yield* bus.publish(McpEvent.StatusChanged, { server: name }).pipe(Effect.ignore)
-        }).pipe(Effect.ensuring(Deferred.succeed(entry.startup, undefined)))
+        }).pipe(Effect.ensuring(entry.startup.open))
 
       const stopServer = Effect.fnUntraced(function* (name: ServerName, entry: ServerEntry) {
         const scope = entry.scope
@@ -562,7 +562,7 @@ export const layer = (options?: Options) =>
         const entry: ServerEntry = {
           config: serverConfig,
           status: { status: "pending" },
-          startup: Deferred.makeUnsafe<void>(),
+          startup: Latch.makeUnsafe(),
         }
         entries.set(name, entry)
         yield* Effect.gen(function* () {
@@ -575,7 +575,7 @@ export const layer = (options?: Options) =>
           yield* startServer(name, entry)
         }).pipe(
           // Settle startup even when registration fails or replacement is interrupted, so readers cannot hang.
-          Effect.ensuring(Effect.sync(() => Deferred.doneUnsafe(entry.startup, Exit.void))),
+          Effect.ensuring(entry.startup.open),
         )
       })
 
@@ -597,7 +597,7 @@ export const layer = (options?: Options) =>
             entries.set(name, {
               config: server,
               status: { status: "pending" },
-              startup: Deferred.makeUnsafe<void>(),
+              startup: Latch.makeUnsafe(),
             })
           }
           yield* Effect.forEach(entries, ([name, entry]) => register(name, entry), { discard: true })
@@ -607,7 +607,7 @@ export const layer = (options?: Options) =>
           for (const [name, entry] of entries) {
             if (entry.config.disabled) {
               entry.status = { status: "disabled" }
-              Deferred.doneUnsafe(entry.startup, Exit.void)
+              entry.startup.openUnsafe()
               yield* bus.publish(McpEvent.StatusChanged, { server: name }).pipe(Effect.ignore)
               continue
             }
@@ -685,7 +685,7 @@ export const layer = (options?: Options) =>
 
       // Suspend so each await sees current entries; a bare Map iterator is exhausted after one run.
       const whenAllReady = Effect.suspend(() =>
-        Effect.forEach(Array.from(entries.values()), (entry) => Deferred.await(entry.startup), {
+        Effect.forEach(Array.from(entries.values()), (entry) => entry.startup.await, {
           concurrency: "unbounded",
           discard: true,
         }),
@@ -734,7 +734,7 @@ export const layer = (options?: Options) =>
         }),
         callTool: Effect.fn("MCP.callTool")(function* (input) {
           const target = yield* requireServer(input.server)
-          yield* Deferred.await(target.entry.startup)
+          yield* target.entry.startup.await
           if (!target.entry.client)
             return yield* new ToolCallError({
               server: target.name,
@@ -773,7 +773,7 @@ export const layer = (options?: Options) =>
         }),
         prompt: Effect.fn("MCP.prompt")(function* (input) {
           const target = yield* requireServer(input.server)
-          yield* Deferred.await(target.entry.startup)
+          yield* target.entry.startup.await
           if (!target.entry.client) return undefined
           const result = yield* target.entry.client
             .prompt({ name: input.name, args: input.args })
@@ -827,7 +827,7 @@ export const layer = (options?: Options) =>
         }),
         readResource: Effect.fn("MCP.readResource")(function* (input) {
           const target = yield* requireServer(input.server)
-          yield* Deferred.await(target.entry.startup)
+          yield* target.entry.startup.await
           if (!target.entry.client) return undefined
           const result = yield* target.entry.client
             .readResource({ uri: input.uri })

packages/core/src/model-resolver.ts

@@ -164,7 +164,6 @@ const resolveCatalogModel = Effect.fn("ModelResolver.resolveCatalogModel")(funct
     ...nativeCredentialSettings(specifier, credential),
     headers: Provider.mergeHeaders(mapping?.headers, resolved.headers),
     body: Provider.mergeOverlay(mapping?.body, resolved.body),
-    limits: { context: resolved.limit.context, input: resolved.limit.input, output: resolved.limit.output },
   }
   return yield* Effect.try({
     try: () => {

packages/core/src/plugin/provider/lmstudio.ts

@@ -70,13 +70,11 @@ export function make(origin = "http://127.0.0.1:1234", interval: Duration.Input
               input: ["text", ...(item.capabilities?.vision ? ["image"] : [])],
               output: ["text"],
             }
-            model.limit = {
-              context:
-                item.loaded_instances.length === 0
-                  ? item.max_context_length
-                  : Math.min(...item.loaded_instances.map((instance) => instance.config.context_length)),
-              output: 0,
-            }
+            const context =
+              item.loaded_instances.length === 0
+                ? item.max_context_length
+                : Math.min(...item.loaded_instances.map((instance) => instance.config.context_length))
+            if (context > 0) model.limit.context = context
           })
         }
       })

packages/core/src/plugin/provider/ollama.ts

@@ -96,13 +96,10 @@ export function make(origin = "http://127.0.0.1:11434", interval: Duration.Input
               input: ["text", ...(item.show.capabilities?.includes("vision") ? ["image"] : [])],
               output: ["text"],
             }
-            model.limit = {
-              context:
-                Object.entries(item.show.model_info ?? {}).flatMap(([key, value]) =>
-                  key.endsWith(".context_length") && typeof value === "number" && value > 0 ? [value] : [],
-                )[0] ?? 0,
-              output: 0,
-            }
+            const context = Object.entries(item.show.model_info ?? {}).flatMap(([key, value]) =>
+              key.endsWith(".context_length") && typeof value === "number" && value > 0 ? [value] : [],
+            )[0]
+            if (context !== undefined) model.limit.context = context
           })
         }
       })

packages/core/src/plugin/provider/vllm.ts

@@ -55,7 +55,8 @@ export function make(origin = "http://127.0.0.1:8000", interval: Duration.Input
             model.name = item.id
             // Tool calling depends on vLLM server flags and parsers that model discovery does not report.
             model.capabilities = { tools: false, input: ["text"], output: ["text"] }
-            model.limit = { context: item.max_model_len ?? 0, output: 0 }
+            if (typeof item.max_model_len === "number" && item.max_model_len > 0)
+              model.limit.context = item.max_model_len
           })
         }
       })

packages/core/src/plugin/supervisor.ts

@@ -3,7 +3,7 @@ export { Service, type Interface } from "./supervisor-service.js"
 
 import type { Plugin as PluginDefinition } from "@opencode-ai/plugin/effect/plugin"
 import { Event } from "@opencode-ai/schema/config"
-import { Cause, Deferred, Effect, Layer, Schema, Stream } from "effect"
+import { Cause, Effect, Latch, Layer, Schema, Stream } from "effect"
 import path from "path"
 import { pathToFileURL } from "url"
 import { ConfigPluginSource } from "../config/plugin/source.js"
@@ -137,7 +137,7 @@ export const layer = Layer.effect(
     const sdk = yield* SdkPlugins.Service
     const sources = yield* ConfigPluginSource.Service
     const bus = yield* Bus.Service
-    const ready = { current: yield* Deferred.make<void>() }
+    const ready = yield* Latch.make()
     let observed = 0
 
     const activate = Effect.fn("PluginSupervisor.activate")(function* () {
@@ -164,7 +164,7 @@ export const layer = Layer.effect(
       Stream.mapEffect(() =>
         Effect.gen(function* () {
           observed++
-          if (yield* Deferred.isDone(ready.current)) ready.current = yield* Deferred.make<void>()
+          yield* ready.close
           return observed
         }),
       ),
@@ -176,12 +176,12 @@ export const layer = Layer.effect(
       Stream.runForEach((target) =>
         Effect.gen(function* () {
           yield* activate()
-          if (observed === target) yield* Deferred.succeed(ready.current, undefined)
+          if (observed === target) yield* ready.open
         }).pipe(Effect.catchCause((cause) => Effect.logError("failed to reload plugins", { cause }))),
       ),
       Effect.forkScoped({ startImmediately: true }),
     )
-    return Service.of({ flush: Effect.suspend(() => Deferred.await(ready.current)) })
+    return Service.of({ flush: ready.await })
   }),
 )
 

packages/core/src/session/compaction.ts

@@ -116,8 +116,19 @@ export interface Interface extends State.Transformable<Draft> {
 
 export class Service extends Context.Service<Service, Interface>()("@opencode/SessionCompaction") {}
 
-const truncate = (value: string) =>
-  value.length <= TOOL_OUTPUT_MAX_CHARS ? value : `${value.slice(0, TOOL_OUTPUT_MAX_CHARS)}\n[truncated]`
+export const truncateToolOutput = (value: string) => {
+  if (value.length <= TOOL_OUTPUT_MAX_CHARS) return value
+  let end = 0
+  for (let count = 0; count < TOOL_OUTPUT_MAX_CHARS && end < value.length; count++) {
+    const code = value.charCodeAt(end)
+    end +=
+      code >= 0xd800 && code <= 0xdbff && value.charCodeAt(end + 1) >= 0xdc00 && value.charCodeAt(end + 1) <= 0xdfff
+        ? 2
+        : 1
+  }
+  if (end === value.length) return value
+  return `${value.slice(0, end)}\n[truncated]`
+}
 
 export const serializeToolContent = (content: SessionMessage.ToolStateCompleted["content"]) =>
   content
@@ -146,7 +157,7 @@ const serialize = (message: SessionMessage.Info) => {
         if (part.state.status === "completed")
           return [
             `[Assistant tool call]: ${part.name}(${input})`,
-            `[Tool result]: ${truncate(serializeToolContent(part.state.content))}`,
+            `[Tool result]: ${truncateToolOutput(serializeToolContent(part.state.content))}`,
           ]
         if (part.state.status === "error")
           return [`[Assistant tool call]: ${part.name}(${input})`, `[Tool error]: ${part.state.error.message}`]
@@ -157,7 +168,8 @@ const serialize = (message: SessionMessage.Info) => {
   if (message.type === "system") return `[System update]: ${message.text}`
   if (message.type === "synthetic") return `[Synthetic context]: ${message.text}`
   if (message.type === "skill") return `[Skill activated: ${message.name}]\n${message.text}`
-  if (message.type === "shell") return `[Shell]: ${message.command}\n${truncate(message.output?.output ?? "")}`
+  if (message.type === "shell")
+    return `[Shell]: ${message.command}\n${truncateToolOutput(message.output?.output ?? "")}`
   return ""
 }
 

packages/core/src/session/runner/llm.ts

@@ -89,7 +89,15 @@ const classifyToolExits = (
     .flatMap((cause) => {
       if (Cause.hasInterrupts(cause)) return []
       const reasons = cause.reasons.flatMap(
-        (reason): Array<Cause.Reason<never>> => (Cause.isFailReason(reason) ? [] : [reason]),
+        (reason): Array<Cause.Reason<never>> =>
+          Cause.isFailReason(reason)
+            ? isDecline(reason.error)
+              ? []
+              : // A typed failure here broke the ExecuteError contract (the per-fiber
+                // `catchTag("Tool.Error")` consumes honest ones). Surfacing it as a defect
+                // keeps it from being dropped, which would leave its call unsettled forever.
+                [Cause.makeDieReason(reason.error)]
+            : [reason],
       )
       return reasons.length > 0 ? [Cause.fromReasons(reasons)] : []
     })

packages/core/src/shell.ts

@@ -1,7 +1,7 @@
 export * as Shell from "./shell.js"
 
 import path from "path"
-import { Context, Deferred, Duration, Effect, Fiber, Layer, Schema, Schedule, Stream } from "effect"
+import { Context, Deferred, Duration, Effect, Fiber, Latch, Layer, Schema, Schedule, Stream } from "effect"
 import { ChildProcess } from "effect/unstable/process"
 import { produce } from "immer"
 import { Shell } from "@opencode-ai/schema/shell"
@@ -286,7 +286,7 @@ const layer = () =>
               sessions.set(id, session)
 
               const stream = createWriteStream(file)
-              const outputDone = Deferred.makeUnsafe<void>()
+              const outputDone = Latch.makeUnsafe()
               const pump = handle.all.pipe(
                 Stream.runForEach((chunk: Uint8Array) =>
                   Effect.sync(() => {
@@ -304,8 +304,8 @@ const layer = () =>
                         stream.end(() => resolve())
                       }),
                   )
-                  yield* Deferred.succeed(outputDone, undefined)
-                }).pipe(Effect.catch(() => Deferred.succeed(outputDone, undefined))),
+                  yield* outputDone.open
+                }).pipe(Effect.catch(() => outputDone.open)),
               )
               yield* Effect.promise(
                 () =>
@@ -324,7 +324,7 @@ const layer = () =>
                     draft.time.completed = Date.now()
                   })
                   yield* beforeWait
-                  yield* Deferred.await(outputDone)
+                  yield* outputDone.await
                   // Resolve waiters with the terminal Info before any retention eviction, so an evicted
                   // session still reports success rather than the removal NotFoundError. This runs before
                   // the timeout-fiber interrupt below, which on the timeout path would otherwise cancel

packages/core/src/tool/plugin/shell.ts

@@ -2,7 +2,6 @@ export * as ShellTool from "./shell.js"
 
 import path from "path"
 import { ToolFailure } from "@opencode-ai/ai"
-import type { Content } from "@opencode-ai/schema/tool"
 import type { Context as PluginContext } from "@opencode-ai/plugin/effect/plugin"
 import { Deferred, Effect, Schema, Scope } from "effect"
 import { Config } from "../../config.js"
@@ -72,10 +71,27 @@ const Output = Schema.Struct({
 
 type Output = typeof Output.Type
 
-const modelOutput = (output: Output): string | undefined => {
-  if (output.status === "running") return BACKGROUND_INSTRUCTION
-  if (output.timeout) return "Command timed out before completion."
-  return `Command exited with code ${output.exit}.`
+const resultMessages = (output: Output) => {
+  const notice = (() => {
+    if (output.status === "running") return BACKGROUND_INSTRUCTION
+    if (output.timeout) return "Command timed out before completion."
+    if (output.exit !== undefined) return `Command exited with code ${output.exit}.`
+  })()
+  return [output.output, ...(notice ? [notice] : [])]
+}
+
+const toolResult = (output: Output) => {
+  return {
+    output,
+    content: resultMessages(output).map((text) => ({ type: "text" as const, text })),
+    metadata: {
+      status: output.status,
+      truncated: output.truncated,
+      ...(output.exit !== undefined ? { exit: output.exit } : {}),
+      ...(output.shellID !== undefined ? { shellID: output.shellID } : {}),
+      ...(output.timeout !== undefined ? { timeout: output.timeout } : {}),
+    },
+  }
 }
 
 export const Plugin = {
@@ -92,32 +108,50 @@ export const Plugin = {
     const notifyWhenDone = Effect.fn("ShellTool.notifyWhenDone")(function* (
       sessionID: SessionSchema.ID,
       id: string,
+      shellID: string,
       command: string,
+      settled: Deferred.Deferred<Output>,
     ) {
       yield* runtime.job.wait({ id: id }).pipe(
-        Effect.flatMap((result) => {
-          const state =
-            result.info?.status === "completed"
-              ? "completed"
-              : result.info?.status === "error"
-                ? "error"
-                : result.info?.status === "cancelled"
-                  ? "cancelled"
-                  : undefined
-          if (state === undefined) return Effect.void
-          const text =
-            state === "completed"
-              ? (result.info!.output ?? "")
+        Effect.flatMap((result) =>
+          Effect.gen(function* () {
+            const info = result.info
+            if (!info) return
+            const state =
+              info.status === "completed"
+                ? "completed"
+                : info.status === "error"
+                  ? "error"
+                  : info.status === "cancelled"
+                    ? "cancelled"
+                    : undefined
+            if (state === undefined) return
+            const output = state === "completed" ? yield* Deferred.await(settled) : undefined
+            const text = output
+              ? resultMessages(output).join("\n\n")
               : state === "error"
-                ? (result.info!.error ?? "Command failed")
+                ? (info.error ?? "Command failed")
                 : "Command cancelled"
-          return runtime.session.synthetic({
-            sessionID,
-            text: `<shell id="${id}" state="${state}" command="${command}">\n${text}\n</shell>`,
-            description: command,
-            metadata: { source: "shell", jobID: id, state },
-          })
-        }),
+            yield* runtime.session.synthetic({
+              sessionID,
+              text: `<shell id="${id}" state="${state}" command="${command}">\n${text}\n</shell>`,
+              description: command,
+              metadata: {
+                source: "shell",
+                jobID: id,
+                shellID,
+                state,
+                ...(output
+                  ? {
+                      truncated: output.truncated,
+                      ...(output.exit !== undefined ? { exit: output.exit } : {}),
+                      ...(output.timeout !== undefined ? { timeout: output.timeout } : {}),
+                    }
+                  : {}),
+              },
+            })
+          }),
+        ),
         Effect.forkIn(scope, { startImmediately: true }),
       )
     })
@@ -268,7 +302,7 @@ export const Plugin = {
 
               if (input.background === true) {
                 yield* runtime.job.background(job.id)
-                yield* notifyWhenDone(context.sessionID, context.id, info.command)
+                yield* notifyWhenDone(context.sessionID, context.id, info.id, info.command, settled)
                 return {
                   output: BACKGROUND_STARTED,
                   shellID: info.id,
@@ -282,7 +316,7 @@ export const Plugin = {
                 .pipe(Effect.onInterrupt(() => runtime.job.cancel(job.id).pipe(Effect.ignore)))
               if (result?.type === "backgrounded") {
                 yield* shell.timeout(info.id, 0)
-                yield* notifyWhenDone(context.sessionID, context.id, info.command)
+                yield* notifyWhenDone(context.sessionID, context.id, info.id, info.command, settled)
                 return {
                   output: BACKGROUND_STARTED,
                   shellID: info.id,
@@ -296,22 +330,7 @@ export const Plugin = {
 
               return yield* Deferred.await(settled)
             }).pipe(
-              Effect.map((output) => {
-                const content: Array<Content> = [{ type: "text", text: output.output }]
-                const model = modelOutput(output)
-                if (model) content.push({ type: "text", text: model })
-                return {
-                  output,
-                  content,
-                  metadata: {
-                    status: output.status,
-                    truncated: output.truncated,
-                    ...("exit" in output && output.exit !== undefined ? { exit: output.exit } : {}),
-                    ...("shellID" in output && output.shellID !== undefined ? { shellID: output.shellID } : {}),
-                    ...("timeout" in output && output.timeout !== undefined ? { timeout: output.timeout } : {}),
-                  },
-                }
-              }),
+              Effect.map(toolResult),
               Effect.mapError(
                 (error) => new ToolFailure({ message: `Unable to execute command: ${input.command}`, error }),
               ),

packages/core/src/tool/runtime.ts

@@ -13,7 +13,20 @@ export const definition = (tool: Tool.Info<any, any>): ToolDefinition => ({
 export const execute = (tool: Tool.Info<any, any>, input: unknown, context: Tool.Context) =>
   Effect.gen(function* () {
     const decoded = yield* decodeInput(tool.input, input)
-    const result = yield* tool.execute(decoded, context)
+    // Tool implementations declare `Tool.Error` but plugins can fail with anything at
+    // runtime. A foreign typed failure would slip past every `catchTag("Tool.Error")`
+    // downstream and leave its call permanently unsettled, so the declared contract is
+    // enforced here at the untrusted boundary. Declines tunnel through as defects and
+    // interrupts are not errors; neither is touched.
+    const result = yield* tool.execute(decoded, context).pipe(
+      Effect.mapError((error: unknown) =>
+        error instanceof Tool.Error
+          ? error
+          : new Tool.Error({
+              message: error instanceof globalThis.Error ? error.message : String(error),
+            }),
+      ),
+    )
     if (tool.output === undefined) {
       if ("output" in result) return yield* Effect.die("Tool result declared output without an output schema")
       return {

packages/core/test/model-resolver.test.ts

@@ -236,7 +236,6 @@ describe("ModelResolver", () => {
         endpoint: { baseURL: "https://openai.example/v1" },
         defaults: {
           headers: { "x-test": "header" },
-          limits: { context: 100, input: 80, output: 20 },
           http: { body: { custom_extension: { enabled: true } } },
         },
       })
@@ -785,7 +784,6 @@ describe("ModelResolver", () => {
                   region: "test",
                   headers: { "x-package": "header" },
                   body: { custom: true },
-                  limits: { context: 100, output: 20 },
                 })
                 return LanguageModel.make({ id: modelID, provider: "package-provider", route: native.route })
               },
@@ -913,7 +911,6 @@ describe("ModelResolver", () => {
                     baseURL: "https://provider.example/v1",
                     headers: { "x-provider": "header" },
                     body: { custom: true },
-                    limits: { context: 100, output: 20 },
                     providerOptions,
                   })
                   return LanguageModel.make({ id: modelID, provider: "native-provider", route: native.route })

packages/core/test/plugin/provider-lmstudio.test.ts

@@ -71,6 +71,13 @@ describe("LMStudioPlugin", () => {
                   max_context_length: 131_072,
                   capabilities: { vision: false, trained_for_tool_use: false },
                 },
+                {
+                  type: "llm",
+                  key: "unknown-context",
+                  display_name: "Unknown Context",
+                  loaded_instances: [],
+                  max_context_length: 0,
+                },
                 {
                   type: "embedding",
                   key: "nomic-embed",
@@ -104,11 +111,14 @@ describe("LMStudioPlugin", () => {
             family: "gemma4",
             name: "Gemma 4 26B A4B",
             capabilities: { tools: true, input: ["text", "image"], output: ["text"] },
-            limit: { context: 16_384, output: 0 },
+            limit: { context: 16_384, output: 32_000 },
           })
           expect(yield* catalog.model.get(providerID, Model.ID.make("deepseek-r1"))).toMatchObject({
             capabilities: { tools: false, input: ["text"], output: ["text"] },
-            limit: { context: 131_072, output: 0 },
+            limit: { context: 131_072, output: 32_000 },
+          })
+          expect(yield* catalog.model.get(providerID, Model.ID.make("unknown-context"))).toMatchObject({
+            limit: { context: 200_000, output: 32_000 },
           })
           expect(yield* catalog.model.get(providerID, Model.ID.make("nomic-embed"))).toBeUndefined()
         }),

packages/core/test/plugin/provider-ollama.test.ts

@@ -54,6 +54,7 @@ describe("OllamaPlugin", () => {
                 return Response.json({
                   models: [
                     summary("gemma3:4b", "gemma-digest", "gemma3"),
+                    summary("unknown-context", "unknown-digest"),
                     summary("nomic-embed", "embed-digest"),
                     summary("removed-model", "removed-digest"),
                   ],
@@ -68,6 +69,8 @@ describe("OllamaPlugin", () => {
                       capabilities: ["completion", "tools", "vision"],
                       model_info: { "gemma3.context_length": 131_072 },
                     }
+                  : body.model === "unknown-context"
+                    ? show({ family: "unknown", capabilities: ["completion"], context: 0 })
                   : show({ family: "nomic-bert", capabilities: ["embedding"], context: 8192 }),
               )
             },
@@ -98,7 +101,10 @@ describe("OllamaPlugin", () => {
             name: "gemma3:4b",
             family: "gemma3",
             capabilities: { tools: true, input: ["text", "image"], output: ["text"] },
-            limit: { context: 131_072, output: 0 },
+            limit: { context: 131_072, output: 32_000 },
+          })
+          expect(yield* catalog.model.get(providerID, Model.ID.make("unknown-context"))).toMatchObject({
+            limit: { context: 200_000, output: 32_000 },
           })
           expect(yield* catalog.model.get(providerID, Model.ID.make("nomic-embed"))).toBeUndefined()
           expect(requests).toContainEqual({ method: "GET", path: "/api/tags" })

packages/core/test/plugin/provider-vllm.test.ts

@@ -63,7 +63,11 @@ describe("VLLMPlugin", () => {
               state.models++
               return Response.json({
                 object: "list",
-                data: [remoteModel("Qwen/Qwen3-Coder", 65_536), remoteModel("foreign-model", 4096, "other")],
+                data: [
+                  remoteModel("Qwen/Qwen3-Coder", 65_536),
+                  remoteModel("unknown-limit", 0),
+                  remoteModel("foreign-model", 4096, "other"),
+                ],
               })
             },
           }),
@@ -97,7 +101,10 @@ describe("VLLMPlugin", () => {
             modelID: "Qwen/Qwen3-Coder",
             name: "Qwen/Qwen3-Coder",
             capabilities: { tools: false, input: ["text"], output: ["text"] },
-            limit: { context: 65_536, output: 0 },
+            limit: { context: 65_536, output: 32_000 },
+          })
+          expect(yield* catalog.model.get(providerID, Model.ID.make("unknown-limit"))).toMatchObject({
+            limit: { context: 200_000, output: 32_000 },
           })
           expect(yield* catalog.model.get(providerID, Model.ID.make("foreign-model"))).toBeUndefined()
         }),

packages/core/test/session-compact.test.ts

@@ -28,7 +28,7 @@ const location = Location.Ref.make({ directory: AbsolutePath.make("/project") })
 const model = LanguageModel.make({
   id: "summary-model",
   provider: "test",
-  route: OpenAIChat.route.with({ limits: { context: 10_000, output: 1_000 } }),
+  route: OpenAIChat.route,
 })
 let requests: LLMRequest[] = []
 const client = Layer.mock(LLMClient.Service)({

packages/core/test/session-compaction.test.ts

@@ -30,7 +30,7 @@ let requests: LLMRequest[] = []
 const model = LanguageModel.make({
   id: "summary-model",
   provider: "test",
-  route: OpenAIChat.route.with({ limits: { context: 10_000, output: 1_000 } }),
+  route: OpenAIChat.route,
 })
 const cost = [
   {
@@ -110,6 +110,13 @@ test("compaction describes tool media without embedding base64", () => {
   expect(serialized).not.toContain(base64)
 })
 
+test("compaction truncation does not split surrogate pairs", () => {
+  const prefix = "a".repeat(1_999)
+
+  expect(SessionCompaction.truncateToolOutput(`${prefix}😀suffix`)).toBe(`${prefix}😀\n[truncated]`)
+  expect(SessionCompaction.truncateToolOutput("😀".repeat(2_000))).toBe("😀".repeat(2_000))
+})
+
 test("compaction prompt requires the checkpoint headings in order", () => {
   const prompt = SessionCompaction.buildPrompt({ context: ["Conversation history"] })
   expect(prompt.match(/^#{2,3} .+$/gm)).toEqual([

packages/core/test/session-model-transport-live.test.ts

@@ -34,13 +34,16 @@ const exchange = (server: WebSocketServerFixture, id: string): WebSocketChannelE
 
 const withServer = <A>(
   options: WebSocketServerOptions,
-  effect: (server: WebSocketServerFixture) => Effect.Effect<A, unknown, SessionModelTransport.Service>,
+  effect: (
+    server: WebSocketServerFixture,
+    constructor: Socket.WebSocketConstructor["Service"],
+  ) => Effect.Effect<A, unknown, SessionModelTransport.Service>,
 ) =>
   Effect.runPromise(
     Effect.gen(function* () {
       const constructor = yield* Socket.WebSocketConstructor
       const server = yield* makeWebSocketServer(options)
-      return yield* effect(server).pipe(
+      return yield* effect(server, constructor).pipe(
         Effect.provide(
           SessionModelTransport.makeLayer({
             open: (input) =>
@@ -293,8 +296,27 @@ describe("SessionModelTransport local WebSocket server", () => {
     )
   })
 
-  // Effect's browser-compatible constructor does not expose upgrade response bodies or headers.
-  // The real 426 fixture therefore pins the observable contract: a not-sent connect failure and one HTTP fallback.
+  test("preserves a real rejected upgrade response", async () => {
+    await withServer({ upgrade: () => false }, (server, constructor) =>
+      Effect.gen(function* () {
+        const error = yield* WebSocketTransport.open({ url: server.url, headers: Headers.empty }).pipe(
+          Effect.provideService(Socket.WebSocketConstructor, constructor),
+          Effect.flip,
+        )
+
+        expect(error.reason).toMatchObject({
+          _tag: "UnknownProvider",
+          status: 426,
+          http: {
+            request: { method: "GET", url: server.url },
+            response: { status: 426, headers: { "x-upgrade-rejected": "true" } },
+            body: "WebSocket upgrade required",
+          },
+        })
+      }),
+    )
+  })
+
   test("falls back once after a real rejected upgrade", async () => {
     let fallbacks = 0
     await withServer({ upgrade: () => false }, (server) =>

packages/core/test/session-runner.test.ts

@@ -2857,11 +2857,17 @@ describe("SessionRunnerLLM", () => {
 
       yield* TestLLM.push(
         TestLLM.stop(
-          LLMEvent.textStart({ id: "commentary", providerMetadata: { openai: { phase: "commentary" } } }),
+          LLMEvent.textStart({
+            id: "commentary",
+            providerMetadata: { openai: { itemId: "msg_commentary", phase: "commentary" } },
+          }),
           LLMEvent.textDelta({ id: "commentary", text: "Checking." }),
           LLMEvent.textEnd({
             id: "commentary",
-            providerMetadata: { openai: { phase: "commentary" }, anthropic: { ignored: true } },
+            providerMetadata: {
+              openai: { itemId: "msg_commentary", phase: "commentary" },
+              anthropic: { ignored: true },
+            },
           }),
         ),
       )
@@ -2872,7 +2878,7 @@ describe("SessionRunnerLLM", () => {
         { type: "user", text: "Check first" },
         {
           type: "assistant",
-          content: [{ type: "text", text: "Checking.", state: { phase: "commentary" } }],
+          content: [{ type: "text", text: "Checking.", state: { itemId: "msg_commentary", phase: "commentary" } }],
         },
       ])
 
@@ -2884,7 +2890,7 @@ describe("SessionRunnerLLM", () => {
         {
           type: "text",
           text: "Checking.",
-          providerMetadata: { openai: { phase: "commentary" } },
+          providerMetadata: { openai: { itemId: "msg_commentary", phase: "commentary" } },
         },
       ])
     }),

packages/core/test/session-title.test.ts

@@ -27,7 +27,7 @@ let requests: LLMRequest[] = []
 const model = LanguageModel.make({
   id: "title-model",
   provider: "test",
-  route: OpenAIChat.route.with({ limits: { context: 10_000, output: 1_000 } }),
+  route: OpenAIChat.route,
 })
 const cost = [
   {

packages/core/test/tool-execute.test.ts

@@ -94,6 +94,24 @@ test("declared outputs cannot bypass validation and raw outputs stay JSON-compat
   )
 })
 
+test("foreign typed failures settle as Tool.Error at the untrusted boundary", async () => {
+  class ForeignFailure extends Schema.TaggedError<ForeignFailure>()("Plugin.ForeignFailure", {
+    message: Schema.String,
+  }) {}
+  const lying: Info = {
+    name: "lying",
+    description: "Fails with a non-Tool.Error typed failure",
+    input: Schema.Struct({}),
+    execute: () => new ForeignFailure({ message: "transport died" }) as never,
+  }
+
+  const exit = await Effect.runPromiseExit(execute(lying, {}, context))
+  expect(exit._tag).toBe("Failure")
+  const error = exit._tag === "Failure" ? exit.cause.reasons.find((reason) => "error" in reason)?.error : undefined
+  expect(error).toBeInstanceOf(Tool.Error)
+  expect((error as Tool.Error).message).toBe("transport died")
+})
+
 test("execute supports callable namespace tools", async () => {
   const callable: Info = {
     name: "admin",

packages/core/test/tool-shell.test.ts

@@ -760,10 +760,52 @@ describe("ShellTool", () => {
             expect((yield* shell.list()).map((info) => info.id)).toContain(id)
             expect((yield* shell.wait(id)).status).toBe("timeout")
             expect((yield* Fiber.join(admitted)).valueOrUndefined?.data.item.payload).toMatchObject({
+              text: expect.stringContaining("Command timed out before completion."),
               description: idleCommand,
               metadata: {
                 source: "shell",
+                shellID,
                 state: "completed",
+                timeout: true,
+                truncated: false,
+              },
+            })
+          }),
+        )
+      },
+      (tmp) => Effect.promise(() => tmp[Symbol.asyncDispose]().then(() => undefined)),
+    ),
+  )
+
+  it.live("preserves a background command's non-zero exit", () =>
+    Effect.acquireUseRelease(
+      Effect.promise(() => tmpdir()),
+      (tmp) => {
+        reset()
+        return withSession(tmp.path, (registry) =>
+          Effect.gen(function* () {
+            const bus = yield* Bus.Service
+            const admitted = yield* bus.subscribe(SessionEvent.InboxEnqueued).pipe(
+              Stream.filter((event) => event.data.sessionID === sessionID && event.data.item.type === "synthetic"),
+              Stream.runHead,
+              Effect.forkScoped({ startImmediately: true }),
+            )
+            const settled = yield* executeTool(
+              registry,
+              call({ command: bodyExitCommand, background: true }, "call-background-nonzero"),
+            )
+            const shellID = settled.metadata?.shellID
+            expect(typeof shellID).toBe("string")
+            expect((yield* Fiber.join(admitted)).valueOrUndefined?.data.item.payload).toMatchObject({
+              text: expect.stringContaining("Command exited with code 7."),
+              description: bodyExitCommand,
+              metadata: {
+                source: "shell",
+                jobID: "call-background-nonzero",
+                shellID,
+                state: "completed",
+                exit: 7,
+                truncated: false,
               },
             })
           }),

packages/desktop/electron.vite.config.ts

@@ -1,6 +1,4 @@
-import { sentryVitePlugin } from "@sentry/vite-plugin"
 import { defineConfig } from "electron-vite"
-import appPlugin from "@opencode-ai/app/vite"
 
 const channel = (() => {
   const raw = process.env.OPENCODE_CHANNEL
@@ -11,9 +9,10 @@ const channel = (() => {
 
 const nodePtyPkg = `@lydell/node-pty-${process.platform}-${process.arch}`
 
+const appPlugin = (await import("@opencode-ai/app/vite")).default
 const sentry =
   process.env.SENTRY_AUTH_TOKEN && process.env.SENTRY_ORG && process.env.SENTRY_PROJECT
-    ? sentryVitePlugin({
+    ? (await import("@sentry/vite-plugin")).sentryVitePlugin({
         authToken: process.env.SENTRY_AUTH_TOKEN,
         org: process.env.SENTRY_ORG,
         project: process.env.SENTRY_PROJECT,
@@ -34,11 +33,12 @@ export default defineConfig({
       "import.meta.env.OPENCODE_CHANNEL": JSON.stringify(channel),
     },
     build: {
-      rollupOptions: {
+      rolldownOptions: {
         input: { index: "src/main/index.ts" },
         // Keep this identical to electron-vite's Node 20.11+ shim. Its regex insertion can
-        // corrupt bundled TypeScript, while a Rollup banner places the shim safely.
+        // corrupt bundled TypeScript, while an output banner places the shim safely.
         output: {
+          format: "es",
           banner: `
 // -- CommonJS Shims --
 import __cjs_mod__ from 'node:module';
@@ -56,13 +56,14 @@ const require = __cjs_mod__.createRequire(import.meta.url);
         enforce: "pre",
         resolveId(s) {
           if (s === "@lydell/node-pty") return nodePtyPkg
+          return undefined
         },
       },
     ],
   },
   preload: {
     build: {
-      rollupOptions: {
+      rolldownOptions: {
         input: { index: "src/preload/index.ts" },
         output: {
           format: "cjs",
@@ -72,6 +73,9 @@ const require = __cjs_mod__.createRequire(import.meta.url);
     },
   },
   renderer: {
+    experimental: {
+      bundledDev: true,
+    },
     define: {
       "import.meta.env.OPENCODE_VERSION": JSON.stringify(process.env.OPENCODE_VERSION),
       "import.meta.env.VITE_OPENCODE_CHANNEL": JSON.stringify(channel),
@@ -81,7 +85,7 @@ const require = __cjs_mod__.createRequire(import.meta.url);
     root: "src/renderer",
     build: {
       sourcemap: true,
-      rollupOptions: {
+      rolldownOptions: {
         input: {
           main: "src/renderer/index.html",
         },

packages/desktop/package.json

@@ -40,7 +40,6 @@
     "@opencode-ai/ui": "workspace:*",
     "@sentry/solid": "catalog:",
     "@sentry/vite-plugin": "catalog:",
-    "@solid-primitives/i18n": "2.2.1",
     "@solid-primitives/storage": "catalog:",
     "@solidjs/meta": "catalog:",
     "@solidjs/router": "0.15.4",
@@ -50,11 +49,11 @@
     "@valibot/to-json-schema": "1.6.0",
     "electron": "42.3.3",
     "electron-builder": "26.15.2",
-    "electron-vite": "^5",
+    "electron-vite": "6.0.0-beta.1",
     "solid-js": "catalog:",
     "sury": "11.0.0-alpha.4",
     "typescript": "~5.6.2",
-    "vite": "catalog:",
+    "vite": "8.2.1",
     "zod-openapi": "5.4.6"
   },
   "optionalDependencies": {

packages/desktop/scripts/dev.ts

@@ -11,14 +11,17 @@ async function main() {
   process.env.OPENCODE_DISABLE_CHANNEL_DB = "0"
   const options = selectOptions()
   if (options.server.type === "build") process.env.OPENCODE_DESKTOP_SERVER_CHANNEL = "local"
+  process.env.OPENCODE_DESKTOP_ISOLATED_SERVER = "1"
   await prepareDesktop()
   await prepareServer(options.server)
   await startDesktop(options.electron)
 }
 
 async function prepareDesktop() {
-  await $`bun run install-electron`
-  await $`bun ./scripts/copy-icons.ts ${process.env.OPENCODE_CHANNEL ?? "dev"}`
+  await Promise.all([
+    $`bun run install-electron`,
+    $`bun ./scripts/copy-icons.ts ${process.env.OPENCODE_CHANNEL ?? "dev"}`,
+  ])
 }
 
 function selectOptions(): DevOptions {
@@ -46,7 +49,6 @@ async function prepareServer(source: ServerSource) {
 }
 
 async function startDesktop(args: string[]) {
-  process.env.OPENCODE_DESKTOP_ISOLATED_SERVER = "1"
   await $`electron-vite dev ${args}`
 }
 

packages/desktop/src/main/index.ts

@@ -2,7 +2,12 @@ import { app } from "electron"
 import { Deferred, Effect, Fiber } from "effect"
 import type { ServerReadyData } from "../shared/ipc-contract"
 import { checkAppExists, resolveAppPath } from "./files/apps"
-import { registerIpcHandlers, registerUpdaterIpcHandlers, registerWslIpcHandlers } from "./ipc"
+import {
+  registerIpcHandlers,
+  registerUpdaterIpcHandlers,
+  registerWslInitialization,
+  registerWslIpcHandlers,
+} from "./ipc"
 import {
   acquireApplicationLock,
   configureApplication,
@@ -26,13 +31,17 @@ const main = Effect.gen(function* () {
   const logger = configureApplication()
   if (!acquireApplicationLock()) return
   preferApplicationEnvironment(logger)
+  loadProxyEnvironment(logger)
   const lifecycle = createApplicationLifecycle(logger)
   const serverReady = Deferred.makeUnsafe<ServerReadyData, unknown>()
+  const wslReady = Promise.withResolvers<void>()
+  logger.log("starting v2 background service")
+  const backgroundTask = yield* Effect.promise(() => startBackgroundCli(logger)).pipe(Effect.forkChild)
 
   yield* Effect.promise(() => app.whenReady())
   yield* prepareDesktop(logger)
 
-  const updater = setupAutoUpdater(lifecycle.prepareToRestart)
+  const updater = yield* Effect.promise(() => setupAutoUpdater(lifecycle.prepareToRestart))
   const menu = {
     trigger: (id: string) => {
       const win = getLastFocusedWindow()
@@ -68,27 +77,35 @@ const main = Effect.gen(function* () {
     },
   })
   registerUpdaterIpcHandlers(createUpdaterIpc(updater))
+  registerWslInitialization(wslReady.promise)
   startAutoUpdater(updater)
   yield* Effect.promise(() => startNetworkLogging())
 
   const loadingTask = yield* Effect.gen(function* () {
-    loadProxyEnvironment(logger)
-    logger.log("starting v2 background service")
-    const background = yield* Effect.promise(() => startBackgroundCli(logger))
-    const wsl = yield* Effect.promise(() => startWsl(background, logger))
-    registerWslIpcHandlers(wsl.ipc)
-    wsl.start()
-    lifecycle.setWslShutdown(wsl.stop)
+    const background = yield* Fiber.join(backgroundTask)
     yield* Deferred.succeed(serverReady, {
       url: background.url,
       username: background.username,
       password: background.password,
     })
     logger.log("loading task finished")
+
+    void startWsl(background, logger).then(
+      (wsl) => {
+        registerWslIpcHandlers(wsl.ipc)
+        lifecycle.setWslShutdown(wsl.stop)
+        wsl.start()
+        wslReady.resolve()
+      },
+      (error) => {
+        logger.error("failed to start WSL manager", { error })
+        wslReady.reject(error)
+      },
+    )
   }).pipe(forwardInitializationFailure(serverReady), Effect.forkChild)
 
-  yield* Fiber.await(loadingTask)
   if (lifecycle.restoreWindows().length) createMenu(menu)
+  yield* Fiber.await(loadingTask)
 })
 
 Effect.runFork(main)