feat(ai): add native Bedrock Mantle support

refactor(tui): remove redundant code (#40081 )
fix(tui): discover plugins across config roots (#39988 )
2026-08-02 08:25:00 -04:00 · 2026-08-02 03:02:03 +00:00 · 2026-08-01 14:22:48 -04:00 · 2026-08-01 14:06:58 -04:00 · 2026-08-01 14:06:13 -04:00 · 2026-08-01 13:10:08 -04:00
201 changed files with 4954 additions and 2729 deletions
@@ -4,6 +4,13 @@
 - The default branch in this repo is `dev`.
 - Local `main` ref may not exist; use `dev` or `origin/dev` for diffs.

+## Live V2 TUI Testing
+
+- Run `bun run dev:live` from a development worktree to test its TUI against the currently elected `opencode2` background server and live sessions.
+- Pass a directory after the script when needed, for example `bun run dev:live /path/to/project`.
+- The script discovers the server with `opencode2 service status`, injects its private local credential from `opencode2 service get password`, and uses the `next` TUI storage channel so tabs and other client-local state match the installed client.
+- Prefer `dev:live` over plain `bun run dev` for this workflow. An implicit managed-service connection may replace the live server when the worktree client version differs; explicit `--server` warns and continues without replacing it.
+
 ## Branch Names

 Use a short branch name of at most three words, separated by hyphens. Do not use slashes or type prefixes such as `feat/` or `fix/`.
@@ -67,6 +67,7 @@
        "@opencode-ai/sdk": "file:vendor/opencode-ai-sdk-1.18.8-dev.tgz",
        "@opencode-ai/session-ui": "workspace:*",
        "@opencode-ai/ui": "workspace:*",
+        "@opencode-ai/util": "workspace:*",
        "@pierre/trees": "1.0.0-beta.4",
        "@sentry/solid": "catalog:",
        "@shikijs/transformers": "3.9.2",
@@ -87,7 +88,6 @@
        "@tanstack/solid-query": "5.91.4",
        "@tanstack/solid-virtual": "catalog:",
        "@thisbeyond/solid-dnd": "0.7.5",
-        "diff": "catalog:",
        "effect": "catalog:",
        "fuzzysort": "catalog:",
        "ghostty-web": "github:anomalyco/ghostty-web#83c0a07b8628b748aed073b232cb4b52a6ca11c1",
@@ -383,7 +383,6 @@
        "@opencode-ai/ai": "workspace:*",
        "@opencode-ai/codemode": "workspace:*",
        "@opencode-ai/effect-drizzle-sqlite": "workspace:*",
-        "@opencode-ai/effect-sqlite-node": "workspace:*",
        "@opencode-ai/plugin": "workspace:*",
        "@opencode-ai/schema": "workspace:*",
        "@opencode-ai/util": "workspace:*",
@@ -404,7 +403,6 @@
        "ignore": "7.0.5",
        "immer": "11.1.4",
        "jsonc-parser": "3.3.1",
-        "semver": "^7.6.3",
        "tree-sitter-bash": "0.25.0",
        "tree-sitter-powershell": "0.25.10",
        "turndown": "7.2.0",
@@ -426,7 +424,6 @@
        "@tsconfig/bun": "catalog:",
        "@types/bun": "catalog:",
        "@types/node": "catalog:",
-        "@types/semver": "catalog:",
        "@types/turndown": "5.0.5",
        "@types/which": "3.0.4",
        "drizzle-kit": "catalog:",
@@ -520,6 +517,7 @@
        "@opencode-ai/core": "workspace:*",
        "@opencode-ai/session-ui": "workspace:*",
        "@opencode-ai/ui": "workspace:*",
+        "@opencode-ai/util": "workspace:*",
        "@pierre/diffs": "catalog:",
        "@solidjs/meta": "catalog:",
        "@solidjs/router": "catalog:",
@@ -8,6 +8,7 @@
  "packageManager": "bun@1.3.14",
  "scripts": {
    "dev": "bun run --cwd packages/cli --conditions=browser src/index.ts",
+    "dev:live": "OPENCODE_TUI_CHANNEL=next OPENCODE_PASSWORD=\"$(opencode2 service get password)\" bun run dev --server \"$(opencode2 service status)\"",
    "dev:desktop": "bun --cwd packages/desktop dev",
    "dev:web": "bun --cwd packages/app dev",
    "dev:console": "ulimit -n 10240 2>/dev/null; bun run --cwd packages/console/app dev",
@@ -60,7 +60,7 @@ Other `aisdk:` packages, including Google Vertex, Azure, and Bedrock, currently
 | `@ai-sdk/google-vertex/xai`       | Vertex Chat / Responses                                        | Partial / usable | Decide Chat/Responses selection for catalog models, add runner mapping and recorded coverage, and review xAI-specific request options.                                                                              |
 | `@ai-sdk/azure`                   | Azure OpenAI Chat/Responses facade                             | Partial          | Map runner/catalog metadata to native Azure, handle resourceName/baseURL/apiVersion variants, add AAD/token auth story, and verify Chat vs Responses deployment selection.                                          |
 | `@ai-sdk/amazon-bedrock`          | Bedrock Converse                                               | Partial          | Add default AWS credential chain/profile support, region/inference-profile model ID handling, provider option parity via `additionalModelRequestFields`, guardrails/performance config, and runner/catalog mapping. |
-| `@ai-sdk/amazon-bedrock/mantle`   | Bedrock Mantle OpenAI-compatible Chat/Responses namespace      | Missing          | Decide native Mantle shape, likely separate from Converse because it uses OpenAI-compatible Chat/Responses semantics over Bedrock. Add package mapping and tests.                                                   |
+| `@ai-sdk/amazon-bedrock/mantle`   | Bedrock Mantle OpenAI-compatible Chat/Responses namespace      | Partial / usable | Add default AWS credential chain/profile support; native catalog mapping currently requires bearer auth or explicit static credentials.                                                                             |

 ## Highest-Risk Gaps

@@ -71,30 +71,31 @@ Other `aisdk:` packages, including Google Vertex, Azure, and Bedrock, currently
 5. Azure is only a provider facade, not a full runtime replacement. Native Azure exists, but the catalog runner does not select it, and token auth/resource variants need review.
 6. Provider option typing is uneven. OpenAI, Anthropic, Gemini, Bedrock, and OpenRouter each expose a small typed subset plus raw HTTP overlays; this is useful but not equivalent to AI SDK provider option coverage.
 7. Structured output is not provider-native yet. `LLM.generateObject` still uses a synthetic tool strategy, while the future design expects native structured output where reliable and tool fallback where needed.
-8. Package/namespace boundaries for the current native loading set are explicit in docs and exports. Other exported provider facades are not catalog package entrypoints until they implement the contract. Vertex xAI still needs catalog API selection; the missing native boundary is Bedrock Mantle.
-9. Recorded coverage is uneven. OpenAI, Anthropic, Gemini, Bedrock Converse, Cloudflare, OpenRouter, and several OpenAI-compatible Chat providers have cassettes. Azure, Vertex, and Mantle need first-class recorded scenarios before switching defaults.
+8. Package/namespace boundaries for the current native loading set are explicit in docs and exports. Other exported provider facades are not catalog package entrypoints until they implement the contract. Vertex xAI still needs catalog API selection.
+9. Recorded coverage is uneven. OpenAI, Anthropic, Gemini, Bedrock Converse, Bedrock Mantle, Cloudflare, OpenRouter, and several OpenAI-compatible Chat providers have cassettes. Azure and Vertex still need first-class recorded scenarios before switching defaults.

 ## Native Namespace Shape

 These are implementation/API slices, not separate npm packages.

-| API slice                     | Package-like entrypoint                                 | Purpose                                                                      |
-| ----------------------------- | ------------------------------------------------------- | ---------------------------------------------------------------------------- |
-| OpenAI Chat                   | `@opencode-ai/ai/providers/openai/chat`                 | OpenAI `/chat/completions` semantics.                                        |
-| OpenAI Responses              | `@opencode-ai/ai/providers/openai/responses`            | OpenAI `/responses` semantics with HTTP/WebSocket selected through settings. |
-| OpenAI-compatible Chat        | `@opencode-ai/ai/providers/openai-compatible`           | Generic OpenAI-compatible `/chat/completions`.                               |
-| Open Responses-compatible     | `@opencode-ai/ai/providers/openai-compatible/responses` | Generic provider-neutral `/responses`.                                       |
-| Anthropic-compatible Messages | `@opencode-ai/ai/providers/anthropic-compatible`        | Generic Anthropic-compatible `/messages`.                                    |
-| Anthropic Messages            | `@opencode-ai/ai/providers/anthropic`                   | Anthropic Messages API.                                                      |
-| Gemini Developer API          | `@opencode-ai/ai/providers/google`                      | Google AI Studio Gemini API.                                                 |
-| Vertex Gemini                 | `@opencode-ai/ai/providers/google-vertex/gemini`        | Vertex Gemini API; `providers/google-vertex` is the default alias.           |
-| Vertex Chat                   | `@opencode-ai/ai/providers/google-vertex/chat`          | Vertex OpenAI-compatible Chat Completions for MaaS models.                   |
-| Vertex Responses              | `@opencode-ai/ai/providers/google-vertex/responses`     | Vertex Open Responses for Grok models.                                       |
-| Vertex Messages               | `@opencode-ai/ai/providers/google-vertex/messages`      | Vertex-hosted Anthropic Messages API.                                        |
-| Bedrock Converse              | `@opencode-ai/ai/providers/amazon-bedrock`              | AWS Bedrock Converse API.                                                    |
-| Bedrock Mantle                | Missing                                                 | AWS Bedrock Mantle OpenAI-compatible APIs.                                   |
-| Azure OpenAI Chat             | `@opencode-ai/ai/providers/azure/chat`                  | Azure specialization of OpenAI Chat.                                         |
-| Azure OpenAI Responses        | `@opencode-ai/ai/providers/azure/responses`             | Azure specialization of OpenAI Responses.                                    |
+| API slice                     | Package-like entrypoint                                     | Purpose                                                                      |
+| ----------------------------- | ----------------------------------------------------------- | ---------------------------------------------------------------------------- |
+| OpenAI Chat                   | `@opencode-ai/ai/providers/openai/chat`                     | OpenAI `/chat/completions` semantics.                                        |
+| OpenAI Responses              | `@opencode-ai/ai/providers/openai/responses`                | OpenAI `/responses` semantics with HTTP/WebSocket selected through settings. |
+| OpenAI-compatible Chat        | `@opencode-ai/ai/providers/openai-compatible`               | Generic OpenAI-compatible `/chat/completions`.                               |
+| Open Responses-compatible     | `@opencode-ai/ai/providers/openai-compatible/responses`     | Generic provider-neutral `/responses`.                                       |
+| Anthropic-compatible Messages | `@opencode-ai/ai/providers/anthropic-compatible`            | Generic Anthropic-compatible `/messages`.                                    |
+| Anthropic Messages            | `@opencode-ai/ai/providers/anthropic`                       | Anthropic Messages API.                                                      |
+| Gemini Developer API          | `@opencode-ai/ai/providers/google`                          | Google AI Studio Gemini API.                                                 |
+| Vertex Gemini                 | `@opencode-ai/ai/providers/google-vertex/gemini`            | Vertex Gemini API; `providers/google-vertex` is the default alias.           |
+| Vertex Chat                   | `@opencode-ai/ai/providers/google-vertex/chat`              | Vertex OpenAI-compatible Chat Completions for MaaS models.                   |
+| Vertex Responses              | `@opencode-ai/ai/providers/google-vertex/responses`         | Vertex Open Responses for Grok models.                                       |
+| Vertex Messages               | `@opencode-ai/ai/providers/google-vertex/messages`          | Vertex-hosted Anthropic Messages API.                                        |
+| Bedrock Converse              | `@opencode-ai/ai/providers/amazon-bedrock`                  | AWS Bedrock Converse API.                                                    |
+| Bedrock Mantle Chat           | `@opencode-ai/ai/providers/amazon-bedrock/mantle/chat`      | AWS Bedrock Mantle OpenAI-compatible Chat API.                               |
+| Bedrock Mantle Responses      | `@opencode-ai/ai/providers/amazon-bedrock/mantle/responses` | AWS Bedrock Mantle OpenAI-compatible Responses API.                          |
+| Azure OpenAI Chat             | `@opencode-ai/ai/providers/azure/chat`                      | Azure specialization of OpenAI Chat.                                         |
+| Azure OpenAI Responses        | `@opencode-ai/ai/providers/azure/responses`                 | Azure specialization of OpenAI Responses.                                    |

 ## Suggested Next Work Slices

@@ -103,6 +104,5 @@ These are implementation/API slices, not separate npm packages.
 3. Bring Bedrock native auth/config to AI SDK parity: region, profile, default AWS credential chain, bearer token env, endpoint override, and cross-region inference profile handling.
 4. Add runner/catalog mappings and recorded scenarios for the native Vertex Gemini, Chat, Responses, and Messages entrypoints.
 5. Decide Chat/Responses selection for `@ai-sdk/google-vertex/xai` catalog models.
-6. Add Bedrock Mantle as a separate OpenAI-compatible Bedrock namespace after deciding whether it uses Chat, Responses, or both by model.
-7. Expand typed provider options from the existing V1 lowerer knowledge in `packages/core/src/v1/config/provider-options.ts` before adding more raw overlay examples.
-8. Add recorded provider tests for Azure, Vertex Gemini, Vertex Chat, Vertex Responses, Vertex Messages, Bedrock credential-chain behavior, and Mantle before making native runtime the default for those packages.
+6. Expand typed provider options from the existing V1 lowerer knowledge in `packages/core/src/v1/config/provider-options.ts` before adding more raw overlay examples.
+7. Add recorded provider tests for Azure, Vertex Gemini, Vertex Chat, Vertex Responses, Vertex Messages, and Bedrock credential-chain behavior before making native runtime the default for those packages.
@@ -38,7 +38,7 @@ const resolve = (policy: CachePolicy | undefined): CachePolicyObject => {
 // Protocols whose wire format ignores inline cache markers (OpenAI's implicit
 // prefix caching, Gemini's implicit + out-of-band CachedContent). Skip the
 // whole policy pass for these — emitting hints would be harmless but pointless.
-const RESPECTS_INLINE_HINTS = new Set(["anthropic-messages", "bedrock-converse"])
+const RESPECTS_INLINE_HINTS = new Set(["anthropic-messages", "bedrock-converse", "openrouter"])

 const makeHint = (ttlSeconds: number | undefined): CacheHint =>
  ttlSeconds !== undefined ? new CacheHint({ type: "ephemeral", ttlSeconds }) : new CacheHint({ type: "ephemeral" })
@@ -133,6 +133,7 @@ const countHints = (request: LLMRequest) =>

 export const applyCachePolicy = (request: LLMRequest): LLMRequest => {
  if (!RESPECTS_INLINE_HINTS.has(request.model.route.id)) return request
+  if (request.model.route.id === "openrouter" && (request.cache === undefined || request.cache === "auto")) return request
  const policy = resolve(request.cache)
  if (!policy.tools && !policy.system && !policy.messages) return request

@@ -11,6 +11,7 @@ import {
  Usage,
  type FinishReason,
  type FinishReasonDetails,
+  type CacheHint,
  type JsonSchema,
  type LLMRequest,
  type MediaPart,
@@ -38,6 +39,11 @@ export const PATH = "/chat/completions"
 // The body schema is the provider-native JSON body. `fromRequest` below builds
 // this shape from the common `LLMRequest`, then `Route.make` validates and
 // JSON-encodes it before transport.
+const OpenAIChatCacheControl = Schema.Struct({
+  type: Schema.Literal("ephemeral"),
+  ttl: Schema.optional(Schema.String),
+})
+
 const OpenAIChatFunction = Schema.Struct({
  name: Schema.String,
  description: Schema.String,
@@ -47,6 +53,7 @@ const OpenAIChatFunction = Schema.Struct({
 const OpenAIChatTool = Schema.Struct({
  type: Schema.tag("function"),
  function: OpenAIChatFunction,
+  cache_control: Schema.optional(OpenAIChatCacheControl),
 })
 type OpenAIChatTool = Schema.Schema.Type<typeof OpenAIChatTool>

@@ -61,7 +68,11 @@ const OpenAIChatAssistantToolCall = Schema.Struct({
 type OpenAIChatAssistantToolCall = Schema.Schema.Type<typeof OpenAIChatAssistantToolCall>

 const OpenAIChatUserContent = Schema.Union([
-  Schema.Struct({ type: Schema.Literal("text"), text: Schema.String }),
+  Schema.Struct({
+    type: Schema.Literal("text"),
+    text: Schema.String,
+    cache_control: Schema.optional(OpenAIChatCacheControl),
+  }),
  Schema.Struct({
    type: Schema.Literal("image_url"),
    image_url: Schema.Struct({ url: Schema.String }),
@@ -69,7 +80,10 @@ const OpenAIChatUserContent = Schema.Union([
 ])

 const OpenAIChatMessage = Schema.Union([
-  Schema.Struct({ role: Schema.Literal("system"), content: Schema.String }),
+  Schema.Struct({
+    role: Schema.Literal("system"),
+    content: Schema.Union([Schema.String, Schema.Array(OpenAIChatUserContent)]),
+  }),
  Schema.Struct({
    role: Schema.Literal("user"),
    content: Schema.Union([Schema.String, Schema.Array(OpenAIChatUserContent)]),
@@ -83,10 +97,16 @@ const OpenAIChatMessage = Schema.Union([
      reasoning: Schema.optional(Schema.String),
      reasoning_text: Schema.optional(Schema.String),
      reasoning_details: Schema.optional(Schema.Unknown),
+      cache_control: Schema.optional(OpenAIChatCacheControl),
    }),
    [Schema.Record(Schema.String, Schema.Unknown)],
  ),
-  Schema.Struct({ role: Schema.Literal("tool"), tool_call_id: Schema.String, content: Schema.String }),
+  Schema.Struct({
+    role: Schema.Literal("tool"),
+    tool_call_id: Schema.String,
+    content: Schema.String,
+    cache_control: Schema.optional(OpenAIChatCacheControl),
+  }),
 ]).pipe(Schema.toTaggedUnion("role"))
 type OpenAIChatMessage = Schema.Schema.Type<typeof OpenAIChatMessage>

@@ -107,6 +127,7 @@ export const bodyFields = {
  stream_options: Schema.optional(Schema.Struct({ include_usage: Schema.Boolean })),
  store: Schema.optional(Schema.Boolean),
  reasoning_effort: Schema.optional(OpenAIOptions.OpenAIReasoningEffort),
+  max_completion_tokens: Schema.optional(Schema.Number),
  max_tokens: Schema.optional(Schema.Number),
  temperature: Schema.optional(Schema.Number),
  top_p: Schema.optional(Schema.Number),
@@ -209,13 +230,20 @@ export interface ParserState {
 // Lowering is the only place that knows how common LLM messages map onto the
 // OpenAI Chat wire format. Keep provider quirks here instead of leaking native
 // fields into `LLMRequest`.
-const lowerTool = (tool: ToolDefinition, inputSchema: JsonSchema): OpenAIChatTool => ({
+interface LoweringOptions {
+  readonly cacheControl?: (
+    cache: CacheHint | undefined,
+  ) => Schema.Schema.Type<typeof OpenAIChatCacheControl> | undefined
+}
+
+const lowerTool = (tool: ToolDefinition, inputSchema: JsonSchema, options: LoweringOptions): OpenAIChatTool => ({
  type: "function",
  function: {
    name: tool.name,
    description: tool.description,
    parameters: ToolSchemaProjection.openAI(inputSchema),
  },
+  cache_control: options.cacheControl?.(tool.cache),
 })

 const lowerToolChoice = (toolChoice: NonNullable<LLMRequest["toolChoice"]>) =>
@@ -257,11 +285,14 @@ const reasoningDetails = (parts: ReadonlyArray<ReasoningPart>, native: unknown)
  if (isRecord(native) && Array.isArray(native.reasoning_details)) return native.reasoning_details
 }

-const lowerUserMessage = Effect.fn("OpenAIChat.lowerUserMessage")(function* (message: OpenAIChatRequestMessage) {
+const lowerUserMessage = Effect.fn("OpenAIChat.lowerUserMessage")(function* (
+  message: OpenAIChatRequestMessage,
+  options: LoweringOptions,
+) {
  const content: Array<Schema.Schema.Type<typeof OpenAIChatUserContent>> = []
  for (const part of message.content) {
    if (part.type === "text") {
-      content.push({ type: "text", text: part.text })
+      content.push({ type: "text", text: part.text, cache_control: options.cacheControl?.(part.cache) })
      continue
    }
    if (part.type === "media") {
@@ -270,14 +301,18 @@ const lowerUserMessage = Effect.fn("OpenAIChat.lowerUserMessage")(function* (mes
    }
    return yield* ProviderShared.unsupportedContent("OpenAI Chat", "user", ["text", "media"])
  }
-  if (content.every((part) => part.type === "text"))
-    return { role: "user" as const, content: content.map((part) => part.text).join("") }
+  if (content.every((part) => part.type === "text" && part.cache_control === undefined))
+    return {
+      role: "user" as const,
+      content: content.map((part) => (part.type === "text" ? part.text : "")).join(""),
+    }
  return { role: "user" as const, content }
 })

 const lowerAssistantMessage = Effect.fn("OpenAIChat.lowerAssistantMessage")(function* (
  message: OpenAIChatRequestMessage,
  configuredField?: string,
+  options: LoweringOptions = {},
 ) {
  const content: TextPart[] = []
  const reasoning: ReasoningPart[] = []
@@ -315,29 +350,44 @@ const lowerAssistantMessage = Effect.fn("OpenAIChat.lowerAssistantMessage")(func
    if (reasoning.length === 0) return nativeReasoning
    return text
  })()
+  const cached = message.content.findLast((part) => "cache" in part && part.cache !== undefined)
  const result = {
    role: "assistant" as const,
    content: content.length === 0 ? null : ProviderShared.joinText(content),
    tool_calls: toolCalls.length === 0 ? undefined : toolCalls,
    reasoning_details: details,
+    cache_control: options.cacheControl?.(cached && "cache" in cached ? cached.cache : undefined),
  }
  if (field === undefined || reasoningText === undefined) return result
  return { ...result, [field]: reasoningText }
 })

-const lowerToolMessages = Effect.fn("OpenAIChat.lowerToolMessages")(function* (message: OpenAIChatRequestMessage) {
+const lowerToolMessages = Effect.fn("OpenAIChat.lowerToolMessages")(function* (
+  message: OpenAIChatRequestMessage,
+  options: LoweringOptions,
+) {
  const messages: OpenAIChatMessage[] = []
  const images: Array<Schema.Schema.Type<typeof OpenAIChatUserContent>> = []
  for (const part of message.content) {
    if (!ProviderShared.supportsContent(part, ["tool-result"]))
      return yield* ProviderShared.unsupportedContent("OpenAI Chat", "tool", ["tool-result"])
    if (part.result.type !== "content") {
-      messages.push({ role: "tool", tool_call_id: part.id, content: ProviderShared.toolResultText(part) })
+      messages.push({
+        role: "tool",
+        tool_call_id: part.id,
+        content: ProviderShared.toolResultText(part),
+        cache_control: options.cacheControl?.(part.cache),
+      })
      continue
    }
    const content: ReadonlyArray<Tool.Content> = part.result.value
    const text = content.filter((item) => item.type === "text").map((item) => item.text)
-    messages.push({ role: "tool", tool_call_id: part.id, content: text.join("\n") })
+    messages.push({
+      role: "tool",
+      tool_call_id: part.id,
+      content: text.join("\n"),
+      cache_control: options.cacheControl?.(part.cache),
+    })
    const files = content.filter((item) => item.type === "file")
    images.push(
      ...(yield* Effect.forEach(files, (item) =>
@@ -351,15 +401,29 @@ const lowerToolMessages = Effect.fn("OpenAIChat.lowerToolMessages")(function* (m
 const lowerMessage = Effect.fn("OpenAIChat.lowerMessage")(function* (
  message: OpenAIChatRequestMessage,
  reasoningField?: string,
+  options: LoweringOptions = {},
 ) {
-  if (message.role === "user") return [yield* lowerUserMessage(message)]
-  if (message.role === "assistant") return [yield* lowerAssistantMessage(message, reasoningField)]
-  return (yield* lowerToolMessages(message)).messages
+  if (message.role === "user") return [yield* lowerUserMessage(message, options)]
+  if (message.role === "assistant") return [yield* lowerAssistantMessage(message, reasoningField, options)]
+  return (yield* lowerToolMessages(message, options)).messages
 })

-const lowerMessages = Effect.fn("OpenAIChat.lowerMessages")(function* (request: LLMRequest) {
+const lowerMessages = Effect.fn("OpenAIChat.lowerMessages")(function* (request: LLMRequest, options: LoweringOptions) {
  const system: OpenAIChatMessage[] =
-    request.system.length === 0 ? [] : [{ role: "system", content: ProviderShared.joinText(request.system) }]
+    request.system.length === 0
+      ? []
+      : request.system.some((part) => part.cache !== undefined) && options.cacheControl !== undefined
+        ? [
+            {
+              role: "system",
+              content: request.system.map((part) => ({
+                type: "text",
+                text: part.text,
+                cache_control: options.cacheControl?.(part.cache),
+              })),
+            },
+          ]
+        : [{ role: "system", content: ProviderShared.joinText(request.system) }]
  const messages = [...system]
  const pendingImages: Array<Schema.Schema.Type<typeof OpenAIChatUserContent>> = []
  const flushImages = () => {
@@ -370,28 +434,53 @@ const lowerMessages = Effect.fn("OpenAIChat.lowerMessages")(function* (request:
    if (message.role === "system") {
      const part = yield* ProviderShared.wrappedSystemUpdate("OpenAI Chat", message)
      if (pendingImages.length > 0) {
-        messages.push({ role: "user", content: [...pendingImages.splice(0), { type: "text", text: part.text }] })
+        messages.push({
+          role: "user",
+          content: [
+            ...pendingImages.splice(0),
+            { type: "text", text: part.text, cache_control: options.cacheControl?.(part.cache) },
+          ],
+        })
        continue
      }
      const previous = messages.at(-1)
      if (previous?.role === "user" && typeof previous.content === "string")
-        messages[messages.length - 1] = { role: "user", content: `${previous.content}\n${part.text}` }
+        messages[messages.length - 1] = options.cacheControl?.(part.cache)
+          ? {
+              role: "user",
+              content: [
+                { type: "text", text: previous.content },
+                { type: "text", text: part.text, cache_control: options.cacheControl(part.cache) },
+              ],
+            }
+          : { role: "user", content: `${previous.content}\n${part.text}` }
      else if (previous?.role === "user" && Array.isArray(previous.content))
        messages[messages.length - 1] = {
          role: "user",
-          content: [...previous.content, { type: "text", text: part.text }],
+          content: [
+            ...previous.content,
+            { type: "text", text: part.text, cache_control: options.cacheControl?.(part.cache) },
+          ],
        }
-      else messages.push({ role: "user", content: part.text })
+      else
+        messages.push(
+          options.cacheControl?.(part.cache)
+            ? {
+                role: "user",
+                content: [{ type: "text", text: part.text, cache_control: options.cacheControl(part.cache) }],
+              }
+            : { role: "user", content: part.text },
+        )
      continue
    }
    if (message.role === "tool") {
-      const lowered = yield* lowerToolMessages(message)
+      const lowered = yield* lowerToolMessages(message, options)
      messages.push(...lowered.messages)
      pendingImages.push(...lowered.images)
      continue
    }
    flushImages()
-    messages.push(...(yield* lowerMessage(message, request.model.compatibility?.reasoningField)))
+    messages.push(...(yield* lowerMessage(message, request.model.compatibility?.reasoningField, options)))
  }
  flushImages()
  return messages
@@ -405,7 +494,10 @@ const lowerOptions = (request: LLMRequest) => {
  }
 }

-const fromRequest = Effect.fn("OpenAIChat.fromRequest")(function* (request: LLMRequest) {
+export const fromRequest = Effect.fn("OpenAIChat.fromRequest")(function* (
+  request: LLMRequest,
+  options: LoweringOptions = {},
+) {
  // `fromRequest` returns the provider body only. Endpoint, auth, framing,
  // validation, and HTTP execution are composed by `Route.make`.
  const reasoningField = request.model.compatibility?.reasoningField
@@ -415,19 +507,26 @@ const fromRequest = Effect.fn("OpenAIChat.fromRequest")(function* (request: LLMR
    )
  const generation = request.generation
  const toolSchemaCompatibility = request.model.compatibility?.toolSchema
+  const maxTokensField = request.model.compatibility?.maxTokensField ?? "max_tokens"
  return {
    model: request.model.id,
-    messages: yield* lowerMessages(request),
+    messages: yield* lowerMessages(request, options),
    tools:
      request.tools.length === 0
        ? undefined
        : request.tools.map((tool) =>
-            lowerTool(tool, ToolSchemaProjection.modelCompatibility(tool.inputSchema, toolSchemaCompatibility)),
+            lowerTool(
+              tool,
+              ToolSchemaProjection.modelCompatibility(tool.inputSchema, toolSchemaCompatibility),
+              options,
+            ),
          ),
    tool_choice: request.toolChoice ? yield* lowerToolChoice(request.toolChoice) : undefined,
    stream: true as const,
    stream_options: { include_usage: true },
-    max_tokens: generation?.maxTokens,
+    ...(maxTokensField === "max_completion_tokens"
+      ? { max_completion_tokens: generation?.maxTokens }
+      : { max_tokens: generation?.maxTokens }),
    temperature: generation?.temperature,
    top_p: generation?.topP,
    frequency_penalty: generation?.frequencyPenalty,
@@ -22,6 +22,8 @@ const signRequest = (input: {
  readonly body: string
  readonly headers: Headers.Headers
  readonly credentials: Credentials
+  readonly service: string
+  readonly name: string
 }) =>
  Effect.tryPromise({
    try: async () => {
@@ -34,23 +36,26 @@ const signRequest = (input: {
        accessKeyId: input.credentials.accessKeyId,
        secretAccessKey: input.credentials.secretAccessKey,
        sessionToken: input.credentials.sessionToken,
-        service: "bedrock",
+        service: input.service,
      }).sign()
      return Object.fromEntries(signed.headers.entries())
    },
    catch: (error) =>
      ProviderShared.invalidRequest(
-        `Bedrock Converse SigV4 signing failed: ${error instanceof Error ? error.message : String(error)}`,
+        `${input.name} SigV4 signing failed: ${error instanceof Error ? error.message : String(error)}`,
      ),
  })

 /** Sign the exact JSON bytes with SigV4 using credentials configured on the route. */
-export const sigV4 = (credentials: Credentials | undefined) =>
+export const sigV4 = (
+  credentials: Credentials | undefined,
+  options: { readonly service?: string; readonly name?: string } = {},
+) =>
  Auth.custom((input: AuthInput) => {
    return Effect.gen(function* () {
      if (!credentials) {
        return yield* ProviderShared.invalidRequest(
-          "Bedrock Converse requires either route bearer auth or AWS credentials configured on the route",
+          `${options.name ?? "Bedrock Converse"} requires either route bearer auth or AWS credentials configured on the route`,
        )
      }
      const headersForSigning = Headers.set(input.headers, "content-type", "application/json")
@@ -59,6 +64,8 @@ export const sigV4 = (credentials: Credentials | undefined) =>
        body: input.body,
        headers: headersForSigning,
        credentials,
+        service: options.service ?? "bedrock",
+        name: options.name ?? "Bedrock Converse",
      })
      return Headers.setAll(headersForSigning, signed)
    })
@@ -16,7 +16,6 @@ import {

 const patterns = [
  /prompt is too long/i,
-  /request_too_large/i,
  /input is too long for requested model/i,
  /exceeds the context window/i,
  /exceeds (?:the )?(?:model'?s )?maximum context length(?: of [\d,]+ tokens?|\s*\([\d,]+\))/i,
@@ -33,7 +32,6 @@ const patterns = [
  /context window exceeds limit/i,
  /exceeded model token limit/i,
  /context[_ ]length[_ ]exceeded/i,
-  /request entity too large/i,
  /context length is only \d+ tokens/i,
  /input length.*exceeds.*context length/i,
  /prompt too long; exceeded (?:max )?context length/i,
@@ -44,11 +42,15 @@ const patterns = [
  /token limit exceeded/i,
 ]

+const payloadPatterns = [/request_too_large/i, /request entity too large/i, /payload too large/i, /request too large/i]
+
 const exclusions = [/^(throttling error|service unavailable):/i, /rate limit/i, /too many requests/i]

 export const isContextOverflow = (message: string) =>
  !exclusions.some((pattern) => pattern.test(message)) &&
-  (patterns.some((pattern) => pattern.test(message)) || /^4(00|13)\s*(status code)?\s*\(no body\)/i.test(message))
+  (patterns.some((pattern) => pattern.test(message)) || /^400\s*(status code)?\s*\(no body\)/i.test(message))
+
+export const isPayloadTooLarge = (message: string) => payloadPatterns.some((pattern) => pattern.test(message))

 export const isContextOverflowFailure = (failure: unknown) =>
  failure instanceof LLMError
@@ -100,6 +102,8 @@ export function classifyProviderFailure(input: ProviderFailure): LLMError["reaso
      isContextOverflow(text))
  )
    return new InvalidRequestReason({ ...common, classification: "context-overflow" })
+  if (input.status === 413 || isPayloadTooLarge(text))
+    return new InvalidRequestReason({ ...common, classification: "payload-too-large" })
  if (CONTENT_POLICY_TEXT.test(text)) return new ContentPolicyReason(common)
  if (codes.some((code) => QUOTA_CODES.has(code)) || (input.status === 429 && QUOTA_TEXT.test(text)))
    return new QuotaExceededReason(common)
@@ -142,12 +146,7 @@ export function classifyProviderFailure(input: ProviderFailure): LLMError["reaso
      retryAfterMs: input.retryAfterMs,
    })
  if (codes.some((code) => INVALID_REQUEST_CODES.has(code))) return new InvalidRequestReason(common)
-  if (
-    input.status === 400 ||
-    input.status === 404 ||
-    input.status === 413 ||
-    input.status === 422
-  )
+  if (input.status === 400 || input.status === 404 || input.status === 413 || input.status === 422)
    return new InvalidRequestReason(common)
  return new UnknownProviderReason({ ...common, status: input.status })
 }
@@ -6,6 +6,7 @@ export interface Settings extends Readonly<Record<string, unknown>> {
  readonly body?: Readonly<Record<string, unknown>>
  readonly limits?: {
    readonly context: number
+    readonly input?: number
    readonly output: number
  }
 }
@@ -0,0 +1,107 @@
+import { Auth } from "../route/auth"
+import type { Route as RouteDef, RouteDefaultsInput } from "../route/client"
+import type { ProviderPackage } from "../provider-package"
+import { OpenAIChat } from "../protocols/openai-chat"
+import { OpenAIResponses } from "../protocols/openai-responses"
+import { BedrockAuth, type Credentials } from "../protocols/utils/bedrock-auth"
+import { ProviderID, type ModelID } from "../schema"
+import { withOpenAIOptions, type OpenAIProviderOptionsInput } from "./openai-options"
+
+export const id = ProviderID.make("amazon-bedrock")
+
+export type Config = RouteDefaultsInput & {
+  readonly apiKey?: string
+  readonly baseURL?: string
+  readonly credentials?: Credentials
+  readonly region?: string
+  readonly providerOptions?: OpenAIProviderOptionsInput
+}
+
+export interface Settings extends ProviderPackage.Settings {
+  readonly apiKey?: string
+  readonly auth?: "bearer" | "sigv4"
+  readonly baseURL?: string
+  readonly credentials?: Credentials
+  readonly region?: string
+  readonly providerOptions?: OpenAIProviderOptionsInput
+}
+
+const responsesRoute = OpenAIResponses.route.with({
+  id: "bedrock-mantle-responses",
+  provider: id,
+})
+
+const chatRoute = OpenAIChat.route.with({
+  id: "bedrock-mantle-chat",
+  provider: id,
+})
+
+export const routes = [responsesRoute, chatRoute]
+
+const configuredRoute = <Body, Prepared>(route: RouteDef<Body, Prepared>, input: Config) => {
+  const region = input.region ?? input.credentials?.region ?? "us-east-1"
+  const credentials = input.credentials === undefined ? undefined : { ...input.credentials, region }
+  return route.with({
+    endpoint: { baseURL: input.baseURL ?? `https://bedrock-mantle.${region}.api.aws/v1` },
+    auth:
+      input.apiKey === undefined
+        ? BedrockAuth.sigV4(credentials, { service: "bedrock-mantle", name: "Bedrock Mantle" })
+        : Auth.bearer(input.apiKey),
+  })
+}
+
+const defaults = (input: Config) => {
+  const { apiKey: _, baseURL: _baseURL, credentials: _credentials, region: _region, ...rest } = input
+  return rest
+}
+
+export const configure = (input: Config = {}) => {
+  const configuredResponsesRoute = configuredRoute(responsesRoute, input)
+  const configuredChatRoute = configuredRoute(chatRoute, input)
+  const modelDefaults = defaults(input)
+  const responses = (modelID: string | ModelID) =>
+    configuredResponsesRoute
+      .with(withOpenAIOptions(modelID, modelDefaults))
+      .model<OpenAIProviderOptionsInput>({ id: modelID })
+  const chat = (modelID: string | ModelID) =>
+    configuredChatRoute
+      .with(withOpenAIOptions(modelID, modelDefaults))
+      .model<OpenAIProviderOptionsInput>({ id: modelID })
+
+  return {
+    id,
+    model: chat,
+    chat,
+    responses,
+    configure,
+  }
+}
+
+export const provider = configure()
+
+const config = (settings: Settings): Config => {
+  if (settings.auth === "bearer" && settings.apiKey === undefined)
+    throw new Error("Amazon Bedrock Mantle bearer auth requires apiKey")
+  if (settings.auth === "sigv4" && settings.apiKey !== undefined)
+    throw new Error("Amazon Bedrock Mantle SigV4 auth does not accept apiKey")
+  return {
+    apiKey: settings.auth === "sigv4" ? undefined : settings.apiKey,
+    baseURL: settings.baseURL,
+    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,
+  }
+}
+
+export const chatModel: ProviderPackage.Definition<Settings, OpenAIProviderOptionsInput>["model"] = (
+  modelID,
+  settings,
+) => configure(config(settings)).chat(modelID)
+export const responsesModel: ProviderPackage.Definition<Settings, OpenAIProviderOptionsInput>["model"] = (
+  modelID,
+  settings,
+) => configure(config(settings)).responses(modelID)
+export const model = chatModel
@@ -0,0 +1,2 @@
+export { chatModel as model } from "../amazon-bedrock-mantle"
+export type { Settings } from "../amazon-bedrock-mantle"
@@ -0,0 +1,2 @@
+export { chatModel as model } from "../../amazon-bedrock-mantle"
+export type { Settings } from "../../amazon-bedrock-mantle"
@@ -0,0 +1,2 @@
+export { responsesModel as model } from "../../amazon-bedrock-mantle"
+export type { Settings } from "../../amazon-bedrock-mantle"
@@ -1,6 +1,7 @@
 export * as Anthropic from "./anthropic"
 export * as AnthropicCompatible from "./anthropic-compatible"
 export * as AmazonBedrock from "./amazon-bedrock"
+export * as AmazonBedrockMantle from "./amazon-bedrock-mantle"
 export * as Azure from "./azure"
 export * as Cloudflare from "./cloudflare"
 export { CloudflareAIGateway, CloudflareWorkersAI } from "./cloudflare"
@@ -4,21 +4,72 @@ import { Endpoint } from "../route/endpoint"
 import { Framing } from "../route/framing"
 import { Protocol } from "../route/protocol"
 import { AuthOptions, type ProviderAuthOption } from "../route/auth-options"
-import { ProviderID, type ModelID, type ProviderOptions } from "../schema"
+import { ProviderID, type CacheHint, type ModelID, type ProviderOptions } from "../schema"
 import type { ProviderPackage } from "../provider-package"
 import * as OpenAICompatibleProfiles from "./openai-compatible-profile"
 import * as OpenAIChat from "../protocols/openai-chat"
+import { newBreakpoints, ttlBucket } from "../protocols/utils/cache"
 import { isRecord } from "../protocols/shared"

 export const profile = OpenAICompatibleProfiles.profiles.openrouter
 export const id = ProviderID.make(profile.provider)
 const ADAPTER = "openrouter"

+type OpenRouterString<Known extends string> = Known | (string & {})
+
+export interface OpenRouterProviderRouting {
+  readonly [key: string]: unknown
+  readonly order?: ReadonlyArray<string>
+  readonly allow_fallbacks?: boolean
+  readonly require_parameters?: boolean
+  readonly data_collection?: OpenRouterString<"allow" | "deny">
+  readonly only?: ReadonlyArray<string>
+  readonly ignore?: ReadonlyArray<string>
+  readonly quantizations?: ReadonlyArray<string>
+  readonly sort?: OpenRouterString<"price" | "throughput" | "latency">
+  readonly max_price?: Readonly<{
+    prompt?: number | string
+    completion?: number | string
+    image?: number | string
+    audio?: number | string
+    request?: number | string
+  }>
+  readonly zdr?: boolean
+}
+
+export type OpenRouterPlugin =
+  | Readonly<{
+      id: "web"
+      max_results?: number
+      search_prompt?: string
+      engine?: OpenRouterString<"native" | "exa">
+    }>
+  | Readonly<{ id: "file-parser"; max_files?: number; pdf?: { engine?: string } }>
+  | Readonly<{ id: "moderation" }>
+  | Readonly<{ id: "response-healing" }>
+  | Readonly<{ id: "auto-router"; allowed_models?: ReadonlyArray<string> }>
+  | Readonly<{ id: string & {}; [key: string]: unknown }>
+
 export interface OpenRouterOptions {
  readonly [key: string]: unknown
-  readonly usage?: boolean | Record<string, unknown>
-  readonly reasoning?: Record<string, unknown>
+  readonly debug?: Readonly<{ echo_upstream_body?: boolean }>
+  readonly models?: ReadonlyArray<string>
+  readonly plugins?: ReadonlyArray<OpenRouterPlugin>
  readonly promptCacheKey?: string
+  readonly provider?: OpenRouterProviderRouting
+  readonly reasoning?: Readonly<{
+    enabled?: boolean
+    exclude?: boolean
+    effort?: OpenRouterString<"none" | "minimal" | "low" | "medium" | "high" | "xhigh" | "max">
+    max_tokens?: number
+  }>
+  readonly usage?: boolean | Readonly<{ include: boolean }>
+  readonly user?: string
+  readonly web_search_options?: Readonly<{
+    max_results?: number
+    search_prompt?: string
+    engine?: OpenRouterString<"native" | "exa">
+  }>
 }

 export type OpenRouterProviderOptionsInput = ProviderOptions & {
@@ -47,7 +98,7 @@ export const protocol = Protocol.make({
  body: {
    schema: OpenRouterBody,
    from: (request) =>
-      OpenAIChat.protocol.body.from(request).pipe(
+      OpenAIChat.fromRequest(request, { cacheControl: cacheControl() }).pipe(
        Effect.map((body) => {
          const sourceAssistants = request.messages.filter((message) => message.role === "assistant")
          let assistantIndex = 0
@@ -78,16 +129,39 @@ export const protocol = Protocol.make({
  stream: OpenAIChat.protocol.stream,
 })

+const cacheControl = () => {
+  const breakpoints = newBreakpoints(4)
+  return (cache: CacheHint | undefined) => {
+    if (cache === undefined || breakpoints.remaining === 0) return undefined
+    breakpoints.remaining -= 1
+    return {
+      type: "ephemeral" as const,
+      ...(ttlBucket(cache.ttlSeconds) === "1h" ? { ttl: "1h" } : {}),
+    }
+  }
+}
+
 const bodyOptions = (input: unknown) => {
  const openrouter = isRecord(input) ? input : {}
+  const { usage, models, provider, plugins, web_search_options, debug, user, reasoning, promptCacheKey, ...options } =
+    openrouter
  return {
-    ...(openrouter.usage === true
+    ...options,
+    ...(usage === undefined || usage === true
      ? { usage: { include: true } }
-      : isRecord(openrouter.usage)
-        ? { usage: openrouter.usage }
-        : {}),
-    ...(isRecord(openrouter.reasoning) ? { reasoning: openrouter.reasoning } : {}),
-    ...(typeof openrouter.promptCacheKey === "string" ? { prompt_cache_key: openrouter.promptCacheKey } : {}),
+      : usage === false
+        ? { usage: { include: false } }
+        : isRecord(usage)
+          ? { usage }
+          : {}),
+    ...(Array.isArray(models) ? { models } : {}),
+    ...(isRecord(provider) ? { provider } : {}),
+    ...(Array.isArray(plugins) ? { plugins } : {}),
+    ...(isRecord(web_search_options) ? { web_search_options } : {}),
+    ...(isRecord(debug) ? { debug } : {}),
+    ...(typeof user === "string" ? { user } : {}),
+    ...(isRecord(reasoning) ? { reasoning } : {}),
+    ...(typeof promptCacheKey === "string" ? { prompt_cache_key: promptCacheKey } : {}),
  }
 }

@@ -28,57 +28,9 @@ const applyQuery = (url: string, query: Record<string, string> | undefined) => {
  return next.toString()
 }

-const PROTOCOL_BODY_OVERLAY_DENYLIST = new Set([
-  "anthropic_version",
-  "content",
-  "contents",
-  "frequencyPenalty",
-  "frequency_penalty",
-  "generationConfig",
-  "inferenceConfig",
-  "input",
-  "maxTokens",
-  "max_tokens",
-  "messages",
-  "model",
-  "presencePenalty",
-  "presence_penalty",
-  "responseFormat",
-  "response_format",
-  "seed",
-  "stop",
-  "stopSequences",
-  "stop_sequences",
-  "stream",
-  "streamOptions",
-  "stream_options",
-  "system",
-  "systemInstruction",
-  "system_instruction",
-  "temperature",
-  "thinking",
-  "toolChoice",
-  "toolConfig",
-  "tool_choice",
-  "tool_config",
-  "tools",
-  "topK",
-  "topP",
-  "top_k",
-  "top_p",
-])
-
-const forbiddenBodyOverlayKeys = (body: Record<string, unknown>) =>
-  Object.keys(body).filter((key) => PROTOCOL_BODY_OVERLAY_DENYLIST.has(key))
-
 const bodyWithOverlay = <Body>(body: Body, request: LLMRequest, encodeBody: (body: Body) => string) =>
  Effect.gen(function* () {
    if (request.http?.body === undefined) return { jsonBody: body, bodyText: encodeBody(body) }
-    const forbiddenKeys = forbiddenBodyOverlayKeys(request.http.body)
-    if (forbiddenKeys.length > 0)
-      return yield* ProviderShared.invalidRequest(
-        `http.body cannot overlay protocol-owned field(s): ${forbiddenKeys.join(", ")}`,
-      )
    if (ProviderShared.isRecord(body)) {
      const overlaid = mergeJsonRecords(body, request.http.body) ?? {}
      return { jsonBody: overlaid, bodyText: ProviderShared.encodeJson(overlaid) }
@@ -2,7 +2,7 @@ import { Schema } from "effect"
 import { Tool } from "@opencode-ai/schema/tool"
 import { ModelID, ProviderID, ProviderMetadata, RouteID } from "./ids"

-export const ProviderFailureClassification = Schema.Literal("context-overflow")
+export const ProviderFailureClassification = Schema.Literals(["context-overflow", "payload-too-large"])
 export type ProviderFailureClassification = typeof ProviderFailureClassification.Type

 export class HttpRequestDetails extends Schema.Class<HttpRequestDetails>("LLM.HttpRequestDetails")({
@@ -124,6 +124,7 @@ export const ToolCallPart = Object.assign(
    name: Schema.String,
    input: Schema.Unknown,
    providerExecuted: Schema.optional(Schema.Boolean),
+    cache: Schema.optional(CacheHint),
    metadata: Schema.optional(Schema.Record(Schema.String, Schema.Unknown)),
    providerMetadata: Schema.optional(ProviderMetadata),
  }).annotate({ identifier: "LLM.Content.ToolCall" }),
@@ -168,6 +169,7 @@ export const ReasoningPart = Schema.Struct({
  type: Schema.Literal("reasoning"),
  text: Schema.String,
  encrypted: Schema.optional(Schema.String),
+  cache: Schema.optional(CacheHint),
  metadata: Schema.optional(Schema.Record(Schema.String, Schema.Unknown)),
  providerMetadata: Schema.optional(ProviderMetadata),
 }).annotate({ identifier: "LLM.Content.Reasoning" })
@@ -123,6 +123,7 @@ export const mergeGenerationOptions = (...items: ReadonlyArray<GenerationOptions

 export class ModelLimits extends Schema.Class<ModelLimits>("LLM.ModelLimits")({
  context: Schema.optional(Schema.Number),
+  input: Schema.optional(Schema.Number),
  output: Schema.optional(Schema.Number),
 }) {}

@@ -166,9 +167,13 @@ export namespace ModelDefaults {
 export const ModelToolSchemaCompatibility = Schema.Literals(["gemini", "moonshot"])
 export type ModelToolSchemaCompatibility = Schema.Schema.Type<typeof ModelToolSchemaCompatibility>

+export const ModelMaxTokensFieldCompatibility = Schema.Literals(["max_completion_tokens", "max_tokens"])
+export type ModelMaxTokensFieldCompatibility = Schema.Schema.Type<typeof ModelMaxTokensFieldCompatibility>
+
 export class ModelCompatibility extends Schema.Class<ModelCompatibility>("LLM.ModelCompatibility")({
  toolSchema: Schema.optional(ModelToolSchemaCompatibility),
  reasoningField: Schema.optional(Schema.String),
+  maxTokensField: Schema.optional(ModelMaxTokensFieldCompatibility),
 }) {}

 export namespace ModelCompatibility {
@@ -171,24 +171,27 @@ describe("request option precedence", () => {
    ),
  )

-  it.effect("rejects raw body overlays for protocol-owned roots", () =>
-    Effect.gen(function* () {
-      const model = OpenAIChat.route
-        .with({ endpoint: { baseURL: "https://api.openai.test/v1/" }, auth: Auth.bearer("test") })
-        .model({ id: "gpt-4o-mini" })
-      const error = yield* compileRequest(
-        LLM.request({
-          model,
-          prompt: "Say hello.",
-          http: { body: { model: "gpt-5", messages: [], tools: [] } },
-        }),
-      ).pipe(Effect.flip)
-
-      expect(error.reason).toMatchObject({
-        _tag: "InvalidRequest",
-        message: "http.body cannot overlay protocol-owned field(s): model, messages, tools",
-      })
-    }),
+  it.effect("applies raw body overlays after protocol lowering", () =>
+    LLMClient.generate(
+      LLM.request({
+        model: OpenAIChat.route
+          .with({ endpoint: { baseURL: "https://api.openai.test/v1/" }, auth: Auth.bearer("test") })
+          .model({ id: "gpt-4o-mini" }),
+        prompt: "Say hello.",
+        http: { body: { model: "gpt-5", messages: [], tools: [] } },
+      }),
+    ).pipe(
+      Effect.provide(
+        dynamicResponse((input) =>
+          Effect.gen(function* () {
+            expect(decodeJson(input.text)).toMatchObject({ model: "gpt-5", messages: [], tools: [] })
+            return input.respond(sseEvents(deltaChunk({}, "stop")), {
+              headers: { "content-type": "text/event-stream" },
+            })
+          }),
+        ),
+      ),
+    ),
  )

  it.effect("uses model output limits after route limits and before call maxTokens", () =>
@@ -85,14 +85,17 @@ describe("RequestExecutor", () => {
    ),
  )

-  it.effect("does not classify generic HTTP 413 payload errors as context overflow", () =>
+  it.effect("classifies generic HTTP 413 payload errors", () =>
    Effect.gen(function* () {
      const executor = yield* RequestExecutor.Service
      const error = yield* executor.execute(request).pipe(Effect.flip)

      expectLLMError(error)
-      expect(error.reason).toMatchObject({ _tag: "InvalidRequest" })
-      expect("classification" in error.reason ? error.reason.classification : undefined).toBeUndefined()
+      expect(error.reason).toMatchObject({
+        _tag: "InvalidRequest",
+        classification: "payload-too-large",
+        http: { response: { status: 413 } },
+      })
    }).pipe(Effect.provide(responsesLayer([new Response("request too large", { status: 413 })]))),
  )

@@ -6,7 +6,6 @@ describe("provider error classification", () => {
  test("classifies provider token limit messages as context overflow", () => {
    const messages = [
      "tokens in request more than max tokens allowed",
-      '{"error":{"type":"request_too_large","message":"Request exceeds the maximum size"}}',
      "Requested token count exceeds the model's maximum context length of 131072 tokens.",
      "Input length (265330) exceeds model's maximum context length (262144).",
      "Input length 131393 exceeds the maximum allowed input length of 131040 tokens.",
@@ -19,6 +18,24 @@ describe("provider error classification", () => {
    expect(messages.every(isContextOverflow)).toBe(true)
  })

+  test("classifies request size failures separately from context overflow", () => {
+    const failures = [
+      classifyProviderFailure({ message: "request too large", status: 413 }),
+      classifyProviderFailure({
+        message: '{"error":{"type":"request_too_large","message":"Request exceeds the maximum size"}}',
+        status: 400,
+      }),
+      classifyProviderFailure({ message: "upstream request entity too large", status: 502 }),
+    ]
+
+    expect(failures).toEqual(
+      failures.map((failure) =>
+        expect.objectContaining({ _tag: "InvalidRequest", classification: "payload-too-large" }),
+      ),
+    )
+    expect(isContextOverflow("413 status code (no body)")).toBe(false)
+  })
+
  test("does not classify rate limits as context overflow", () => {
    const messages = [
      "Throttling error: Too many tokens, please wait before trying again.",
@@ -59,9 +76,10 @@ describe("provider error classification", () => {
  })

  test("classifies transient client statuses as provider internal", () => {
-    expect(
-      [408, 409].map((status) => classifyProviderFailure({ message: `HTTP ${status}`, status })._tag),
-    ).toEqual(["ProviderInternal", "ProviderInternal"])
+    expect([408, 409].map((status) => classifyProviderFailure({ message: `HTTP ${status}`, status })._tag)).toEqual([
+      "ProviderInternal",
+      "ProviderInternal",
+    ])
  })

  test("classifies nested provider codes when a top-level code is also present", () => {
@@ -5,6 +5,28 @@ const model = OpenRouter.provider.model("anthropic/claude-sonnet-4.5")

 LLM.request({ model, prompt: "Hello", providerOptions: { openrouter: { usage: true } } })

+LLM.request({
+  model,
+  prompt: "Hello",
+  providerOptions: {
+    openrouter: {
+      models: ["google/gemini-3.1-pro"],
+      provider: {
+        order: ["anthropic"],
+        require_parameters: true,
+        data_collection: "future-policy",
+        sort: "future-sort",
+        max_price: { prompt: "0.50" },
+      },
+      reasoning: { effort: "future-effort", exclude: false },
+      plugins: [{ id: "future-plugin", enabled: true }],
+      web_search_options: { engine: "future-engine" },
+      debug: { echo_upstream_body: true },
+      user: "user_123",
+    },
+  },
+})
+
 LLM.request({
  model,
  prompt: "Hello",
@@ -23,12 +23,16 @@ describe("provider package entrypoints", () => {
      import("@opencode-ai/ai/providers/google-vertex/messages"),
      import("@opencode-ai/ai/providers/openrouter"),
      import("@opencode-ai/ai/providers/xai"),
+      import("@opencode-ai/ai/providers/amazon-bedrock/mantle"),
+      import("@opencode-ai/ai/providers/amazon-bedrock/mantle/chat"),
+      import("@opencode-ai/ai/providers/amazon-bedrock/mantle/responses"),
    ])

    for (const module of modules) expect(module.model).toBeFunction()
    expect(modules[0].model).toBe(modules[1].model)
    expect(modules[8].model).toBe(modules[9].model)
    expect(modules[12].model).toBe(modules[13].model)
+    expect(modules[19].model).toBe(modules[20].model)
  })

  test("maps OpenRouter and xAI package settings onto executable models", async () => {
@@ -0,0 +1,111 @@
+import { describe, expect } from "bun:test"
+import { Effect } from "effect"
+import { HttpClientRequest } from "effect/unstable/http"
+import { LLM } from "../../src"
+import { AmazonBedrockMantle } from "../../src/providers"
+import { compileRequest, LLMClient } from "../../src/route/client"
+import { it } from "../lib/effect"
+import { dynamicResponse } from "../lib/http"
+import { recordedTests } from "../recorded-test"
+
+const credentials = {
+  region: "us-east-2",
+  accessKeyId: "AKIAIOSFODNN7EXAMPLE",
+  secretAccessKey: "wJalrXUtnFEMI/K7MDENG/bPxRfiCYEXAMPLEKEY",
+}
+
+describe("Amazon Bedrock Mantle provider", () => {
+  it.effect("uses Chat by default and exposes Responses", () =>
+    Effect.gen(function* () {
+      const provider = AmazonBedrockMantle.configure({ credentials })
+      const chat = yield* compileRequest(LLM.request({ model: provider.model("openai.gpt-oss-120b"), prompt: "Hi" }))
+      const responses = yield* compileRequest(
+        LLM.request({ model: provider.responses("openai.gpt-oss-120b"), prompt: "Hi" }),
+      )
+
+      expect(chat).toMatchObject({
+        route: "bedrock-mantle-chat",
+        protocol: "openai-chat",
+        body: { model: "openai.gpt-oss-120b" },
+      })
+      expect(responses).toMatchObject({
+        route: "bedrock-mantle-responses",
+        protocol: "openai-responses",
+        body: { model: "openai.gpt-oss-120b", store: false },
+      })
+    }),
+  )
+
+  it.effect("uses the Mantle endpoint and signing service", () =>
+    Effect.gen(function* () {
+      const seen: Array<{ readonly url: string; readonly authorization: string | undefined }> = []
+      const model = AmazonBedrockMantle.configure({ credentials, region: "us-west-1" }).responses("openai.gpt-oss-120b")
+      yield* LLMClient.generate(LLM.request({ model, prompt: "Hi" })).pipe(
+        Effect.provide(
+          dynamicResponse((input) =>
+            Effect.gen(function* () {
+              const request = yield* HttpClientRequest.toWeb(input.request)
+              seen.push({ url: request.url, authorization: request.headers.get("authorization") ?? undefined })
+              return input.respond("", { headers: { "content-type": "text/event-stream" } })
+            }),
+          ),
+        ),
+        Effect.flip,
+      )
+
+      expect(seen[0]?.url).toBe("https://bedrock-mantle.us-west-1.api.aws/v1/responses")
+      expect(seen[0]?.authorization).toContain("/us-west-1/bedrock-mantle/aws4_request")
+    }),
+  )
+
+  it.effect("supports bearer authentication and custom base URLs", () =>
+    Effect.gen(function* () {
+      const seen: Array<{ readonly url: string; readonly authorization: string | undefined }> = []
+      const model = AmazonBedrockMantle.configure({
+        apiKey: "test-key",
+        baseURL: "https://mantle.test/v1",
+      }).chat("openai.gpt-oss-safeguard-20b")
+      yield* LLMClient.generate(LLM.request({ model, prompt: "Hi" })).pipe(
+        Effect.provide(
+          dynamicResponse((input) =>
+            Effect.gen(function* () {
+              const request = yield* HttpClientRequest.toWeb(input.request)
+              seen.push({ url: request.url, authorization: request.headers.get("authorization") ?? undefined })
+              return input.respond("", { headers: { "content-type": "text/event-stream" } })
+            }),
+          ),
+        ),
+        Effect.flip,
+      )
+
+      expect(seen).toEqual([{ url: "https://mantle.test/v1/chat/completions", authorization: "Bearer test-key" }])
+    }),
+  )
+})
+
+const recorded = recordedTests({
+  prefix: "bedrock-mantle",
+  provider: "amazon-bedrock",
+  protocol: "openai-responses",
+  requires: ["AWS_BEARER_TOKEN_BEDROCK"],
+  metadata: { model: "openai.gpt-oss-120b" },
+})
+
+describe("Amazon Bedrock Mantle recorded", () => {
+  recorded.effect("streams text", () =>
+    Effect.gen(function* () {
+      const response = yield* LLMClient.generate(
+        LLM.request({
+          model: AmazonBedrockMantle.configure({
+            apiKey: process.env.AWS_BEARER_TOKEN_BEDROCK ?? "fixture",
+            region: "us-east-1",
+          }).responses("openai.gpt-oss-120b"),
+          prompt: "Reply with exactly: hello",
+          generation: { maxTokens: 256, temperature: 0 },
+        }),
+      )
+
+      expect(response.text.trim().toLowerCase()).toBe("hello")
+    }),
+  )
+})
@@ -181,23 +181,6 @@ describe("Google Vertex providers", () => {
    }),
  )

-  it.effect("protects the Vertex Messages API version from body overlays", () =>
-    Effect.gen(function* () {
-      const error = yield* compileRequest(
-        LLM.request({
-          model: GoogleVertexMessages.configure({
-            accessToken: "vertex-token",
-            http: { body: { anthropic_version: "wrong" } },
-            project: "vertex-project",
-          }).model("claude-sonnet-4-6"),
-          prompt: "Say hello.",
-        }),
-      ).pipe(Effect.flip)
-
-      expect(error.message).toContain("http.body cannot overlay protocol-owned field(s): anthropic_version")
-    }),
-  )
-
  it.effect("routes tuned Gemini models through their deployed endpoint", () =>
    Effect.gen(function* () {
      const response = yield* LLMClient.generate(
@@ -144,6 +144,20 @@ describe("OpenAI-compatible Chat route", () => {
    }),
  )

+  it.effect("configures the max tokens request field", () =>
+    Effect.gen(function* () {
+      const compatible = OpenAICompatibleChat.route
+        .with({ provider: "custom", endpoint: { baseURL: "https://api.custom.test/v1" } })
+        .model({ id: "custom-model", compatibility: { maxTokensField: "max_completion_tokens" } })
+      const prepared = yield* compileRequest(
+        LLM.request({ model: compatible, prompt: "Say hello.", generation: { maxTokens: 20 } }),
+      )
+
+      expect(prepared.body).toMatchObject({ max_completion_tokens: 20 })
+      expect(prepared.body).not.toHaveProperty("max_tokens")
+    }),
+  )
+
  it.effect("matches AI SDK compatible tool request body fixture", () =>
    Effect.gen(function* () {
      const prepared = yield* compileRequest(
@@ -1,6 +1,6 @@
 import { describe, expect } from "bun:test"
 import { Effect } from "effect"
-import { LLM, Message } from "../../src"
+import { CacheHint, LLM, Message } from "../../src"
 import { LLMClient } from "../../src/route"
 import { compileRequest } from "../../src/route/client"
 import * as OpenRouter from "../../src/providers/openrouter"
@@ -27,10 +27,131 @@ describe("OpenRouter", () => {
        model: "openai/gpt-4o-mini",
        messages: [{ role: "user", content: "Say hello." }],
        stream: true,
+        usage: { include: true },
      })
    }),
  )

+  it.effect("lowers the native cache policy to OpenRouter cache controls", () =>
+    Effect.gen(function* () {
+      const prepared = yield* compileRequest(
+        LLM.request({
+          model: OpenRouter.configure({ apiKey: "test-key" }).model("anthropic/claude-sonnet-4.6"),
+          system: [
+            { type: "text", text: "Base agent", cache: new CacheHint({ type: "ephemeral", ttlSeconds: 3_600 }) },
+            { type: "text", text: "Project instructions" },
+          ],
+          tools: [{ name: "lookup", description: "Lookup", inputSchema: { type: "object", properties: {} } }],
+          prompt: "Hello",
+          cache: { tools: true, system: true, messages: { tail: 1 } },
+        }),
+      )
+
+      expect(prepared.body).toMatchObject({
+        tools: [{ cache_control: { type: "ephemeral" } }],
+        messages: [
+          {
+            role: "system",
+            content: [
+              { text: "Base agent", cache_control: { type: "ephemeral", ttl: "1h" } },
+              { text: "Project instructions", cache_control: { type: "ephemeral" } },
+            ],
+          },
+          {
+            role: "user",
+            content: [{ text: "Hello", cache_control: { type: "ephemeral" } }],
+          },
+        ],
+      })
+    }),
+  )
+
+  it.effect("lowers manual assistant and tool-result cache hints", () =>
+    Effect.gen(function* () {
+      const prepared = yield* compileRequest(
+        LLM.request({
+          model: OpenRouter.configure({ apiKey: "test-key" }).model("anthropic/claude-sonnet-4.6"),
+          cache: "none",
+          messages: [
+            Message.user("Call the tool"),
+            Message.assistant([
+              { type: "text", text: "Calling", cache: new CacheHint({ type: "ephemeral" }) },
+              { type: "tool-call", id: "call_1", name: "lookup", input: {} },
+            ]),
+            Message.tool({
+              id: "call_1",
+              name: "lookup",
+              result: "Done",
+              cache: new CacheHint({ type: "ephemeral", ttlSeconds: 3_600 }),
+            }),
+          ],
+        }),
+      )
+
+      expect(prepared.body.messages).toMatchObject([
+        { role: "user", content: "Call the tool" },
+        { role: "assistant", content: "Calling", cache_control: { type: "ephemeral" } },
+        { role: "tool", content: '"Done"', cache_control: { type: "ephemeral", ttl: "1h" } },
+      ])
+    }),
+  )
+
+  it.effect("caps manual cache controls at four breakpoints", () =>
+    Effect.gen(function* () {
+      const cache = new CacheHint({ type: "ephemeral" })
+      const prepared = yield* compileRequest(
+        LLM.request({
+          model: OpenRouter.configure({ apiKey: "test-key" }).model("anthropic/claude-sonnet-4.6"),
+          cache: "none",
+          system: [1, 2, 3, 4, 5].map((index) => ({ type: "text" as const, text: `System ${index}`, cache })),
+          prompt: "Hello",
+        }),
+      )
+
+      const system = prepared.body.messages[0]
+      expect(system?.role).toBe("system")
+      expect(
+        system && Array.isArray(system.content)
+          ? system.content.filter((part) => "cache_control" in part && part.cache_control !== undefined)
+          : [],
+      ).toHaveLength(4)
+    }),
+  )
+
+  it.effect("preserves cache policy hints on reasoning-only assistant messages", () =>
+    Effect.gen(function* () {
+      const prepared = yield* compileRequest(
+        LLM.request({
+          model: OpenRouter.configure({ apiKey: "test-key" }).model("anthropic/claude-sonnet-4.6"),
+          cache: { messages: "latest-assistant" },
+          messages: [Message.user("Think"), Message.assistant([{ type: "reasoning", text: "Reasoning" }])],
+        }),
+      )
+
+      expect(prepared.body.messages).toMatchObject([
+        { role: "user", content: "Think" },
+        { role: "assistant", cache_control: { type: "ephemeral" } },
+      ])
+    }),
+  )
+
+  it.effect("allows usage accounting to be disabled explicitly", () =>
+    Effect.gen(function* () {
+      const prepared = yield* compileRequest(
+        LLM.request({
+          model: OpenRouter.configure({
+            apiKey: "test-key",
+            providerOptions: { openrouter: { usage: false } },
+          }).model("openai/gpt-4o-mini"),
+          cache: "none",
+          prompt: "Hello",
+        }),
+      )
+
+      expect(prepared.body.usage).toEqual({ include: false })
+    }),
+  )
+
  it.effect("applies OpenRouter payload options from the model helper", () =>
    Effect.gen(function* () {
      const prepared = yield* compileRequest(
@@ -42,6 +163,13 @@ describe("OpenRouter", () => {
                usage: true,
                reasoning: { effort: "high" },
                promptCacheKey: "session_123",
+                models: ["anthropic/claude-sonnet-4.6", "google/gemini-3.1-pro"],
+                provider: { order: ["anthropic", "google"], require_parameters: true },
+                plugins: [{ id: "response-healing" }],
+                web_search_options: { engine: "native", max_results: 3 },
+                debug: { echo_upstream_body: true },
+                user: "user_123",
+                future_option: { enabled: true },
              },
            },
          }).model("anthropic/claude-3.7-sonnet:thinking"),
@@ -53,10 +181,54 @@ describe("OpenRouter", () => {
        usage: { include: true },
        reasoning: { effort: "high" },
        prompt_cache_key: "session_123",
+        models: ["anthropic/claude-sonnet-4.6", "google/gemini-3.1-pro"],
+        provider: { order: ["anthropic", "google"], require_parameters: true },
+        plugins: [{ id: "response-healing" }],
+        web_search_options: { engine: "native", max_results: 3 },
+        debug: { echo_upstream_body: true },
+        user: "user_123",
+        future_option: { enabled: true },
      })
    }),
  )

+  it.effect("filters invalid known OpenRouter options while preserving extensions", () =>
+    Effect.gen(function* () {
+      const invalid: Record<string, unknown> = {
+        usage: "yes",
+        models: "anthropic/claude-sonnet-4.6",
+        provider: [],
+        plugins: {},
+        web_search_options: [],
+        debug: [],
+        user: 123,
+        reasoning: [],
+        promptCacheKey: 123,
+        future_option: { enabled: true },
+      }
+      const prepared = yield* compileRequest(
+        LLM.request({
+          model: OpenRouter.configure({
+            apiKey: "test-key",
+            providerOptions: { openrouter: invalid },
+          }).model("openai/gpt-4o-mini"),
+          prompt: "Hello",
+        }),
+      )
+
+      expect(prepared.body).toMatchObject({ future_option: { enabled: true } })
+      expect(prepared.body).not.toHaveProperty("usage")
+      expect(prepared.body).not.toHaveProperty("models")
+      expect(prepared.body).not.toHaveProperty("provider")
+      expect(prepared.body).not.toHaveProperty("plugins")
+      expect(prepared.body).not.toHaveProperty("web_search_options")
+      expect(prepared.body).not.toHaveProperty("debug")
+      expect(prepared.body).not.toHaveProperty("user")
+      expect(prepared.body).not.toHaveProperty("reasoning")
+      expect(prepared.body).not.toHaveProperty("prompt_cache_key")
+    }),
+  )
+
  it.effect("preserves the upstream provider finish reason", () =>
    Effect.gen(function* () {
      const model = OpenRouter.configure({ apiKey: "test-key" }).model("anthropic/claude-sonnet-4.6")
@@ -104,6 +276,7 @@ describe("OpenRouter", () => {
      const prepared = yield* compileRequest(
        LLM.request({
          model: OpenRouter.configure({ apiKey: "test-key" }).model("anthropic/claude-sonnet-4.6"),
+          cache: "none",
          messages: [
            Message.assistant([
              {
@@ -137,6 +310,7 @@ describe("OpenRouter", () => {
      const prepared = yield* compileRequest(
        LLM.request({
          model: OpenRouter.configure({ apiKey: "test-key" }).model("anthropic/claude-sonnet-4.6"),
+          cache: "none",
          messages: [
            Message.assistant({
              type: "reasoning",
@@ -162,6 +336,7 @@ describe("OpenRouter", () => {
      const prepared = yield* compileRequest(
        LLM.request({
          model: OpenRouter.configure({ apiKey: "test-key" }).model("anthropic/claude-sonnet-4.6"),
+          cache: "none",
          messages: [
            Message.assistant({
              type: "reasoning",
@@ -183,6 +358,7 @@ describe("OpenRouter", () => {
      const prepared = yield* compileRequest(
        LLM.request({
          model: OpenRouter.configure({ apiKey: "test-key" }).model("anthropic/claude-sonnet-4.6"),
+          cache: "none",
          messages: [Message.assistant({ type: "reasoning", text: "Thinking" })],
        }),
      )
@@ -59,6 +59,7 @@
    "@opencode-ai/sdk": "file:vendor/opencode-ai-sdk-1.18.8-dev.tgz",
    "@opencode-ai/session-ui": "workspace:*",
    "@opencode-ai/ui": "workspace:*",
+    "@opencode-ai/util": "workspace:*",
    "@pierre/trees": "1.0.0-beta.4",
    "@sentry/solid": "catalog:",
    "@shikijs/transformers": "3.9.2",
@@ -4,6 +4,7 @@ import { createResizeObserver } from "@solid-primitives/resize-observer"
 import { createMutation } from "@tanstack/solid-query"
 import { IconButtonV2 } from "@opencode-ai/ui/v2/icon-button-v2"
 import { Icon as IconV2 } from "@opencode-ai/ui/v2/icon"
+import { displayLabel } from "@opencode-ai/util/session-title-fallback"
 import { useGlobal } from "@/context/global"
 import { ServerConnection, serverName } from "@/context/server"
 import { displayName, projectForSession } from "@/pages/layout/helpers"
@@ -11,7 +12,6 @@ import { SessionTabAvatar } from "@/pages/layout/session-tab-avatar"
 import type { Session } from "@opencode-ai/sdk/v2"
 import { canOpenTabRename, forwardTabRef } from "./titlebar-tab-gesture"
 import { TabPreviewPopover } from "./titlebar-tab-popover"
-import { sessionTitle } from "@/utils/session-title"
 import "./titlebar-tab-nav.css"

 // MouseEvent.button uses 1 for the middle/wheel button.
@@ -57,7 +57,7 @@ export function TabNavItem(props: {
  })
  const title = createMemo(() => {
    const session = props.session()
-    return session ? sessionTitle(session.title, session.parentID) : props.fallbackTitle
+    return session ? displayLabel(session) : props.fallbackTitle
  })

  const projectName = createMemo(() => {
@@ -3,11 +3,11 @@ import { useLanguage } from "@/context/language"
 import { serverName } from "@/context/server"
 import { displayName } from "@/pages/layout/helpers"
 import { makeEventListener } from "@solid-primitives/event-listener"
+import { displayLabel } from "@opencode-ai/util/session-title-fallback"
 import { createMemo, onCleanup } from "solid-js"
 import { createStore } from "solid-js/store"
 import type { HomeController } from "./home-controller"
 import { homeSessionSearchKey, type HomeSessionRecord, type HomeSessionsController } from "./home-sessions-controller"
-import { sessionTitle } from "@/utils/session-title"

 type HomeSessionSearchSource = Pick<HomeSessionsController, "data" | "session">

@@ -24,7 +24,7 @@ export function createHomeSessionSearchController(home: HomeController, sessions
    if (!value) return []
    return sessions.data
      .searchRecords()
-      .filter((record) => `${sessionTitle(record.session.title)} ${record.projectName}`.toLowerCase().includes(value))
+      .filter((record) => `${displayLabel(record.session)} ${record.projectName}`.toLowerCase().includes(value))
  })
  const active = createMemo(() => {
    const records = results()
@@ -6,10 +6,10 @@ import { ButtonV2 } from "@opencode-ai/ui/v2/button-v2"
 import { Icon as IconV2 } from "@opencode-ai/ui/v2/icon"
 import { IconButtonV2 } from "@opencode-ai/ui/v2/icon-button-v2"
 import { TooltipV2 } from "@opencode-ai/ui/v2/tooltip-v2"
+import { displayLabel } from "@opencode-ai/util/session-title-fallback"
 import { useLanguage } from "@/context/language"
 import { ServerConnection } from "@/context/server"
 import { SessionTabAvatarView } from "@/pages/layout/session-tab-avatar"
-import { sessionTitle } from "@/utils/session-title"
 import { shouldOpenSessionInBackground } from "../home-session-open"
 import {
  HomeSessionStatusController,
@@ -344,7 +344,7 @@ function HomeSessionSearchResultRow(
    selected: boolean
  },
 ) {
-  const title = createMemo(() => sessionTitle(props.record.session.title) || props.record.session.id)
+  const title = createMemo(() => displayLabel(props.record.session))
  const showProjectName = () => props.showProjectName() && props.record.projectName
  const key = () => homeSessionSearchKey(props.record)

@@ -415,7 +415,7 @@ function HomeSessionGroupHeader(props: {
 }

 function HomeSessionRow(props: HomeSessionsViewProps & { record: HomeSessionRecord }) {
-  const title = createMemo(() => sessionTitle(props.record.session.title) || props.record.session.id)
+  const title = createMemo(() => displayLabel(props.record.session))
  const showProjectName = () => props.showProjectName() && props.record.projectName

  return (
@@ -5,6 +5,7 @@ import { Icon as IconV2 } from "@opencode-ai/ui/v2/icon"
 import { IconButton } from "@opencode-ai/ui/icon-button"
 import { Spinner } from "@opencode-ai/ui/spinner"
 import { Tooltip } from "@opencode-ai/ui/tooltip"
+import { displayLabel } from "@opencode-ai/util/session-title-fallback"
 import { getFilename } from "@opencode-ai/core/util/path"
 import { A, useParams } from "@solidjs/router"
 import { type Accessor, createMemo, For, type JSX, Match, Show, Switch } from "solid-js"
@@ -14,7 +15,6 @@ import { getAvatarColors, type LocalProject, useLayout } from "@/context/layout"
 import { useNotification } from "@/context/notification"
 import { usePermission } from "@/context/permission"
 import { messageAgentColor } from "@/utils/agent"
-import { sessionTitle } from "@/utils/session-title"
 import { sessionPermissionRequest } from "../session/composer/session-request-tree"
 import { childSessionOnPath, getProjectAvatarSource, hasProjectPermissions } from "./helpers"

@@ -104,7 +104,7 @@ const SessionRow = (props: {
  warmPress: () => void
  warmFocus: () => void
 }): JSX.Element => {
-  const title = () => sessionTitle(props.session.title, props.session.parentID)
+  const title = () => displayLabel(props.session)

  return (
    <A
@@ -229,7 +229,7 @@ export const SessionItem = (props: SessionItemProps): JSX.Element => {
              fallback={
                <Tooltip
                  placement={props.mobile ? "bottom" : "right"}
-                  value={sessionTitle(props.session.title, props.session.parentID)}
+                  value={displayLabel(props.session)}
                  gutter={10}
                  class="min-w-0 w-full"
                >
@@ -70,7 +70,7 @@ import { legacySessionHref, requireServerKey, sessionHref } from "@/utils/sessio
 import { useSDK } from "@/context/sdk"
 import { useSync } from "@/context/sync"
 import { notifySessionTabsRemoved } from "@/components/titlebar-session-events"
-import { sessionTitle } from "@/utils/session-title"
+import { displayLabel } from "@opencode-ai/util/session-title-fallback"
 import { scheduleConnectedMeasure } from "./measure"
 import { observeElementOffsetReconnectAware } from "./observe-element-offset"
 import { createTimelineProjection } from "./projection"
@@ -296,11 +296,10 @@ export function MessageTimeline(props: {
    if (!id) return
    return sync().session.get(id)
  })
-  const titleValue = createMemo(() => info()?.title)
  const titleLabel = createMemo(() => {
    const session = info()
    if (!session) return
-    return sessionTitle(titleValue(), session.parentID)
+    return displayLabel(session)
  })
  const shareUrl = createMemo(() => info()?.share?.url)
  const shareEnabled = createMemo(() => sync().data.config.share !== "disabled")
@@ -317,7 +316,7 @@ export function MessageTimeline(props: {
  })
  const parentTitle = createMemo(() => {
    const session = parent()
-    return session ? sessionTitle(session.title, session.parentID) : language.t("command.session.new")
+    return session ? displayLabel(session) : language.t("command.session.new")
  })
  const getMsgParts = (msgId: string) => sync().data.part[msgId] ?? emptyParts
  const getMsgPart = (messageID: string, partID: string) => getMsgParts(messageID).find((part) => part.id === partID)
@@ -921,7 +920,7 @@ export function MessageTimeline(props: {
  function DialogDeleteSession(props: { sessionID: string }) {
    const name = createMemo(() => {
      const session = sync().session.get(props.sessionID)
-      return session ? sessionTitle(session.title, session.parentID) : language.t("command.session.new")
+      return session ? displayLabel(session) : language.t("command.session.new")
    })
    const handleDelete = async () => {
      await deleteSession(props.sessionID)
@@ -38,17 +38,19 @@ test("reports a divergent native offset once and ignores equal offsets and unrel
    instance.scrollOffset = offset
  })

-  document.body.append(unrelated)
-  unrelated.remove()
-  await frames(2)
-  expect(calls).toEqual([])
-
  route.remove()
  document.body.append(route)
  await new Promise((resolve) => setTimeout(resolve, 0))
  await frames(3)
  expect(calls).toEqual([[0, false]])

+  instance.scrollOffset = 79_400
+  document.body.append(unrelated)
+  unrelated.remove()
+  await frames(2)
+  expect(calls).toEqual([[0, false]])
+
+  instance.scrollOffset = 0
  route.remove()
  document.body.append(route)
  await new Promise((resolve) => setTimeout(resolve, 0))
@@ -1,11 +0,0 @@
-import { describe, expect, test } from "bun:test"
-import { sessionTitle } from "./session-title"
-
-describe("sessionTitle", () => {
-  test("uses a display fallback without persisting it", () => {
-    expect(sessionTitle(undefined)).toBe("New session")
-    expect(sessionTitle(undefined, "ses_parent")).toBe("Child session")
-    expect(sessionTitle("New session - 2026-07-30T18:45:03.662Z")).toBe("New session")
-    expect(sessionTitle("Generated title")).toBe("Generated title")
-  })
-})
@@ -1,7 +0,0 @@
-const pattern = /^(New session|Child session) - \d{4}-\d{2}-\d{2}T\d{2}:\d{2}:\d{2}\.\d{3}Z$/
-
-export function sessionTitle(title?: string, parentID?: string) {
-  if (!title) return parentID ? "Child session" : "New session"
-  const match = title.match(pattern)
-  return match?.[1] ?? title
-}
@@ -1,5 +1,6 @@
 import type { SessionApi, SessionInfo, SessionListInput } from "@opencode-ai/client/promise"
 import type { Session } from "@opencode-ai/sdk/v2/client"
+import { withTimestampedFallback } from "@opencode-ai/util/session-title-fallback"

 export function normalizeSessionInfo(input: SessionInfo | Session): Session {
  if (!("location" in input)) return input
@@ -13,7 +14,7 @@ export function normalizeSessionInfo(input: SessionInfo | Session): Session {
    parentID: input.parentID,
    cost: input.cost,
    tokens: input.tokens,
-    title: input.title ?? `${input.parentID ? "Child" : "New"} session - ${new Date(input.time.created).toISOString()}`,
+    title: withTimestampedFallback(input),
    agent: input.agent,
    model: input.model,
    version: "",
@@ -58,8 +58,9 @@ export async function collectNodeAssets(target: NodeTarget) {
        source: path.join(ptyRoot, relative),
      })),
  ]
-  await Promise.all(assets.map((asset) => stat(asset.source)))
-  return assets
+  const unique = [...new Map(assets.map((asset) => [asset.key, asset])).values()]
+  await Promise.all(unique.map((asset) => stat(asset.source)))
+  return unique
 }

 export async function hashNodeAssets(assets: readonly NodeAsset[]) {
@@ -8,6 +8,7 @@ import {
  type SessionMessageInfo,
  type SkillInfo,
 } from "@opencode-ai/client/promise"
+import { withTimestampedFallback } from "@opencode-ai/util/session-title-fallback"
 import type {
  AgentSideConnection,
  AuthenticateRequest,
@@ -213,9 +214,7 @@ export function make(input: { readonly client: OpenCodeClient; readonly connecti
        sessions: page.data.map((session) => ({
          sessionId: session.id,
          cwd: session.location.directory,
-          title:
-            session.title ??
-            `${session.parentID ? "Child" : "New"} session - ${new Date(session.time.created).toISOString()}`,
+          title: withTimestampedFallback(session),
          updatedAt: new Date(session.time.updated).toISOString(),
        })),
        ...(page.cursor.next ? { nextCursor: page.cursor.next } : {}),
@@ -133,7 +133,6 @@ export const Commands = Spec.make(typeof OPENCODE_CLI_NAME === "string" ? OPENCO
      description: "Manage plugins",
      commands: [Spec.make("list", { description: "List active plugins" })],
    }),
-    Spec.make("migrate", { description: "Migrate v1 data to v2" }),
    Spec.make("mini", {
      description: "Start the minimal interactive interface",
      params: {
@@ -45,7 +45,11 @@ export default Runtime.handler(Commands, (input) =>
    const runPromise = Effect.runPromiseWith(context)
    const service = server.service
    yield* run({
-      app: { name: process.env.OPENCODE_CLIENT ?? "cli", version: OPENCODE_VERSION, channel: OPENCODE_CHANNEL },
+      app: {
+        name: process.env.OPENCODE_CLIENT ?? "cli",
+        version: OPENCODE_VERSION,
+        channel: process.env.OPENCODE_TUI_CHANNEL ?? OPENCODE_CHANNEL,
+      },
      server: {
        endpoint: server.endpoint,
        service: service
@@ -1,5 +0,0 @@
-import { Effect } from "effect"
-import { Commands } from "../commands"
-import { Runtime } from "../../framework/runtime"
-
-export default Runtime.handler(Commands.commands.migrate, (_input) => Effect.log("No migrations to run."))
@@ -116,13 +116,6 @@ export function migrateV1(legacy: TuiConfigV1.Info | undefined, kv: Record<strin
              : { grouping: kv.exploration_grouping ? ("auto" as const) : ("none" as const) }),
          },
        }),
-    ...(kv.dismissed_getting_started === undefined
-      ? {}
-      : {
-          hints: {
-            onboarding: !kv.dismissed_getting_started,
-          },
-        }),
    ...(kv.animations_enabled === undefined ? {} : { animations: kv.animations_enabled }),
    ...(legacy?.mouse === undefined ? {} : { mouse: legacy.mouse }),
  }
@@ -35,7 +35,6 @@ const Handlers = Runtime.handlers(Commands, {
  plugin: {
    list: () => import("./commands/handlers/plugin/list"),
  },
-  migrate: () => import("./commands/handlers/migrate"),
  mini: () => import("./commands/handlers/mini"),
  run: () => import("./commands/handlers/run"),
  pair: () => import("./commands/handlers/pair"),
@@ -235,10 +235,6 @@ async function renderToolError(part: SessionMessageAssistantTool, directory: str
  UI.println(UI.Style.TEXT_NORMAL + "✗", UI.Style.TEXT_NORMAL + `${info.title} failed`)
 }

-function warning(message: string) {
-  UI.println(UI.Style.TEXT_WARNING_BOLD + "!", UI.Style.TEXT_NORMAL, message)
-}
-
 function errorMessage(error: unknown) {
  if (error instanceof Error) return error.message
  if (typeof error === "object" && error !== null && "message" in error && typeof error.message === "string")
@@ -74,12 +74,12 @@ test("migrates tui and kv config into cli.json", async () => {
      terminal: { title: false },
      prompt: { editor: false, paste: "full" },
      session: { sidebar: "hide", scrollbar: true, thinking: "show", grouping: "none" },
-      hints: { onboarding: false },
      animations: false,
      mouse: false,
    })
    expect(config).not.toHaveProperty("skipped_version")
    expect(config).not.toHaveProperty("which_key")
+    expect(config).not.toHaveProperty("hints")
    expect((await Bun.file(path.join(directory, "cli.json")).json()).keybinds).toEqual({ leader: "ctrl+o" })
    expect(await Bun.file(path.join(directory, "cli.json")).exists()).toBe(true)
    expect(await Bun.file(path.join(directory, "tui.json")).exists()).toBe(true)
@@ -0,0 +1,17 @@
+import { expect, test } from "bun:test"
+import { fileURLToPath } from "node:url"
+import { collectNodeAssets } from "../script/node-assets"
+import { nodeTarget, shellParserWasmAssets } from "../src/node/target"
+
+test("collects each SEA asset key once", async () => {
+  const assets = await collectNodeAssets(nodeTarget(process.platform, process.arch))
+  const keys = assets.map((asset) => asset.key)
+
+  expect(new Set(keys).size).toBe(keys.length)
+  expect(assets.filter((asset) => asset.key === shellParserWasmAssets.runtime)).toEqual([
+    {
+      key: shellParserWasmAssets.runtime,
+      source: fileURLToPath(import.meta.resolve(shellParserWasmAssets.runtime)),
+    },
+  ])
+})
@@ -118,6 +118,7 @@ export type SessionListOperation<E = never> = (input?: Endpoint5_0Input) => Effe

 export type Endpoint5_1Input = {
  readonly id?: Session.ID | undefined
+  readonly title?: string | undefined
  readonly agent?: Agent.ID | undefined
  readonly model?: Model.Ref | undefined
  readonly location?: Location.Ref | undefined
@@ -397,7 +398,7 @@ export type Endpoint5_26Output =
          readonly location?: Location.Ref | undefined
          readonly data: {
            readonly sessionID: Session.ID
-            readonly error: { readonly type: string; readonly message: string }
+            readonly error: { readonly type: string; readonly message: string; readonly status?: number | undefined }
          }
        }
      | {
@@ -523,7 +524,7 @@ export type Endpoint5_26Output =
          readonly data: {
            readonly sessionID: Session.ID
            readonly assistantMessageID: SessionMessage.ID
-            readonly error: { readonly type: string; readonly message: string }
+            readonly error: { readonly type: string; readonly message: string; readonly status?: number | undefined }
            readonly cost?: (number & Brand.Brand<"Money.USD">) | undefined
            readonly tokens?:
              | {
@@ -685,7 +686,7 @@ export type Endpoint5_26Output =
            readonly sessionID: Session.ID
            readonly assistantMessageID: SessionMessage.ID
            readonly callID: string
-            readonly error: { readonly type: string; readonly message: string }
+            readonly error: { readonly type: string; readonly message: string; readonly status?: number | undefined }
            readonly content?:
              | readonly [
                  (
@@ -725,7 +726,7 @@ export type Endpoint5_26Output =
            readonly assistantMessageID: SessionMessage.ID
            readonly attempt: number
            readonly at: number
-            readonly error: { readonly type: string; readonly message: string }
+            readonly error: { readonly type: string; readonly message: string; readonly status?: number | undefined }
          }
        }
      | {
@@ -775,7 +776,7 @@ export type Endpoint5_26Output =
          readonly data: {
            readonly sessionID: Session.ID
            readonly reason: "auto" | "manual"
-            readonly error: { readonly type: string; readonly message: string }
+            readonly error: { readonly type: string; readonly message: string; readonly status?: number | undefined }
            readonly inputID?: SessionMessage.ID | undefined
          }
        }
@@ -299,7 +299,13 @@ const Endpoint5_0 = (raw: RawClient["server.session"]) => (input?: Endpoint5_0In
 const Endpoint5_1 = (raw: RawClient["server.session"]) => (input?: Endpoint5_1Input) =>
  preserveEffect<Endpoint5_1Output>()(
    raw["session.create"]({
-      payload: { id: input?.["id"], agent: input?.["agent"], model: input?.["model"], location: input?.["location"] },
+      payload: {
+        id: input?.["id"],
+        title: input?.["title"],
+        agent: input?.["agent"],
+        model: input?.["model"],
+        location: input?.["location"],
+      },
    }).pipe(
      Effect.mapError(mapClientError),
      Effect.map((value) => value.data),
@@ -462,6 +462,7 @@ export function make(options: ClientOptions) {
            path: `/api/session`,
            body: {
              id: input?.["id"],
+              title: input?.["title"],
              agent: input?.["agent"],
              model: input?.["model"],
              location: input?.["location"],
@@ -108,7 +108,7 @@ export type ToolTextContent = { type: "text"; text: string }

 export type ToolFileContent = { type: "file"; uri: string; mime: string; name?: string | null }

-export type SessionStructuredError = { type: string; message: string }
+export type SessionStructuredError = { type: string; message: string; status?: number }

 export type SessionMessageCompactionRunning = {
  type: "compaction"
@@ -2662,24 +2662,35 @@ export type SessionListOutput = SessionsResponse
 export type SessionCreateInput = {
  readonly id?: {
    readonly id?: string | null
+    readonly title?: string | null
    readonly agent?: string | null
    readonly model?: { readonly id: string; readonly providerID: string; readonly variant?: string } | null
    readonly location?: { readonly directory: string; readonly workspaceID?: string } | null
  }["id"]
+  readonly title?: {
+    readonly id?: string | null
+    readonly title?: string | null
+    readonly agent?: string | null
+    readonly model?: { readonly id: string; readonly providerID: string; readonly variant?: string } | null
+    readonly location?: { readonly directory: string; readonly workspaceID?: string } | null
+  }["title"]
  readonly agent?: {
    readonly id?: string | null
+    readonly title?: string | null
    readonly agent?: string | null
    readonly model?: { readonly id: string; readonly providerID: string; readonly variant?: string } | null
    readonly location?: { readonly directory: string; readonly workspaceID?: string } | null
  }["agent"]
  readonly model?: {
    readonly id?: string | null
+    readonly title?: string | null
    readonly agent?: string | null
    readonly model?: { readonly id: string; readonly providerID: string; readonly variant?: string } | null
    readonly location?: { readonly directory: string; readonly workspaceID?: string } | null
  }["model"]
  readonly location?: {
    readonly id?: string | null
+    readonly title?: string | null
    readonly agent?: string | null
    readonly model?: { readonly id: string; readonly providerID: string; readonly variant?: string } | null
    readonly location?: { readonly directory: string; readonly workspaceID?: string } | null
@@ -56,7 +56,6 @@
    "@tsconfig/bun": "catalog:",
    "@types/bun": "catalog:",
    "@types/node": "catalog:",
-    "@types/semver": "catalog:",
    "@types/turndown": "5.0.5",
    "@types/which": "3.0.4",
    "@parcel/watcher-darwin-arm64": "2.5.1",
@@ -100,7 +99,6 @@
    "@ff-labs/fff-node": "0.10.1",
    "@opencode-ai/codemode": "workspace:*",
    "@opencode-ai/effect-drizzle-sqlite": "workspace:*",
-    "@opencode-ai/effect-sqlite-node": "workspace:*",
    "@opencode-ai/ai": "workspace:*",
    "@opencode-ai/schema": "workspace:*",
    "@opencode-ai/plugin": "workspace:*",
@@ -122,7 +120,6 @@
    "immer": "11.1.4",
    "ignore": "7.0.5",
    "jsonc-parser": "3.3.1",
-    "semver": "^7.6.3",
    "turndown": "7.2.0",
    "tree-sitter-bash": "0.25.0",
    "tree-sitter-powershell": "0.25.10",
@@ -1,15 +1,24 @@
 export * as AISDKNative from "./aisdk-native"

 import { isRecord } from "@opencode-ai/ai/utils/record"
+import { Provider } from "./provider"

 export interface Mapping {
  readonly package: string
  readonly settings: Readonly<Record<string, unknown>>
+  readonly headers?: Readonly<Record<string, string>>
+  readonly body?: Readonly<Record<string, unknown>>
 }

-export function map(packageName: string | undefined, settings: Readonly<Record<string, unknown>>): Mapping | undefined {
+export function map(
+  packageName: string | undefined,
+  settings: Readonly<Record<string, unknown>>,
+  modelID?: string,
+): Mapping | undefined {
  const baseSettings = mapBaseSettings(settings)
  switch (packageName) {
+    case "@ai-sdk/amazon-bedrock/mantle":
+      return mapBedrockMantle(settings, baseSettings, modelID)
    case "@ai-sdk/google":
      return {
        package: "@opencode-ai/ai/providers/google",
@@ -20,14 +29,7 @@ export function map(packageName: string | undefined, settings: Readonly<Record<s
        },
      }
    case "@openrouter/ai-sdk-provider":
-      return {
-        package: "@opencode-ai/ai/providers/openrouter",
-        settings: {
-          ...baseSettings,
-          ...mapAPIKey(settings),
-          ...mapProviderOptions("openrouter", settings),
-        },
-      }
+      return mapOpenRouter(settings, baseSettings)
    case "@ai-sdk/xai":
      return {
        package: "@opencode-ai/ai/providers/xai",
@@ -40,6 +42,79 @@ export function map(packageName: string | undefined, settings: Readonly<Record<s
  }
 }

+function mapBedrockMantle(
+  settings: Readonly<Record<string, unknown>>,
+  baseSettings: Readonly<Record<string, unknown>>,
+  modelID: string | undefined,
+): Mapping | undefined {
+  const apiKey =
+    typeof settings.apiKey === "string"
+      ? settings.apiKey
+      : typeof settings.bearerToken === "string"
+        ? settings.bearerToken
+        : undefined
+  const credentials = mapBedrockCredentials(settings)
+  if (apiKey === undefined && credentials === undefined) return undefined
+  const safeguard = modelID === "openai.gpt-oss-safeguard-20b" || modelID === "openai.gpt-oss-safeguard-120b"
+  return {
+    package: `@opencode-ai/ai/providers/amazon-bedrock/mantle/${safeguard ? "chat" : "responses"}`,
+    settings: {
+      ...baseSettings,
+      ...(typeof settings.baseURL !== "string" && typeof settings.endpoint === "string"
+        ? { baseURL: settings.endpoint }
+        : {}),
+      ...(apiKey === undefined ? {} : { apiKey }),
+      ...(credentials === undefined ? {} : { credentials }),
+      ...(typeof settings.region === "string" ? { region: settings.region } : {}),
+      ...mapOpenAIOptions(settings),
+    },
+  }
+}
+
+function mapBedrockCredentials(settings: Readonly<Record<string, unknown>>) {
+  if (
+    isRecord(settings.credentials) &&
+    typeof settings.credentials.region === "string" &&
+    typeof settings.credentials.accessKeyId === "string" &&
+    typeof settings.credentials.secretAccessKey === "string"
+  ) {
+    return {
+      region: settings.credentials.region,
+      accessKeyId: settings.credentials.accessKeyId,
+      secretAccessKey: settings.credentials.secretAccessKey,
+      ...(typeof settings.credentials.sessionToken === "string"
+        ? { sessionToken: settings.credentials.sessionToken }
+        : {}),
+    }
+  }
+  if (
+    typeof settings.region !== "string" ||
+    typeof settings.accessKeyId !== "string" ||
+    typeof settings.secretAccessKey !== "string"
+  )
+    return undefined
+  return {
+    region: settings.region,
+    accessKeyId: settings.accessKeyId,
+    secretAccessKey: settings.secretAccessKey,
+    ...(typeof settings.sessionToken === "string" ? { sessionToken: settings.sessionToken } : {}),
+  }
+}
+
+function mapOpenAIOptions(settings: Readonly<Record<string, unknown>>) {
+  const options = {
+    ...(typeof settings.reasoningEffort === "string" ? { reasoningEffort: settings.reasoningEffort } : {}),
+    ...(typeof settings.reasoningSummary === "string" ? { reasoningSummary: settings.reasoningSummary } : {}),
+    ...(Array.isArray(settings.include) ? { include: settings.include } : {}),
+    ...(typeof settings.store === "boolean" ? { store: settings.store } : {}),
+    ...(typeof settings.promptCacheKey === "string" ? { promptCacheKey: settings.promptCacheKey } : {}),
+    ...(typeof settings.textVerbosity === "string" ? { textVerbosity: settings.textVerbosity } : {}),
+    ...(typeof settings.serviceTier === "string" ? { serviceTier: settings.serviceTier } : {}),
+  }
+  if (Object.keys(options).length === 0) return {}
+  return { providerOptions: { openai: options } }
+}
+
 function mapBaseSettings(settings: Readonly<Record<string, unknown>>) {
  return {
    ...(typeof settings.baseURL === "string" ? { baseURL: settings.baseURL } : {}),
@@ -69,6 +144,61 @@ function mapGoogleOptions(settings: Readonly<Record<string, unknown>>) {
  return { providerOptions: { gemini: options } }
 }

+function mapOpenRouter(
+  settings: Readonly<Record<string, unknown>>,
+  baseSettings: Readonly<Record<string, unknown>>,
+): Mapping {
+  const headers =
+    Provider.mergeHeaders(
+      {
+        ...(typeof settings.appName === "string" ? { "X-OpenRouter-Title": settings.appName } : {}),
+        ...(typeof settings.appUrl === "string" ? { "HTTP-Referer": settings.appUrl } : {}),
+        ...(isStringRecord(settings.api_keys) && Object.keys(settings.api_keys).length > 0
+          ? { "X-Provider-API-Keys": JSON.stringify(settings.api_keys) }
+          : {}),
+      },
+      isStringRecord(settings.headers) ? settings.headers : undefined,
+    ) ?? {}
+  return {
+    package: "@opencode-ai/ai/providers/openrouter",
+    settings: {
+      ...baseSettings,
+      ...mapAPIKey(settings),
+      ...mapOpenRouterOptions(settings),
+    },
+    ...(Object.keys(headers).length > 0 ? { headers } : {}),
+    ...(isRecord(settings.extraBody) ? { body: settings.extraBody } : {}),
+  }
+}
+
+function mapOpenRouterOptions(settings: Readonly<Record<string, unknown>>) {
+  const options = Object.fromEntries(
+    Object.entries(settings).filter(
+      ([key]) =>
+        ![
+          "apiKey",
+          "api_keys",
+          "appName",
+          "appUrl",
+          "authToken",
+          "baseURL",
+          "chunkTimeout",
+          "compatibility",
+          "extraBody",
+          "fetch",
+          "headers",
+          "timeout",
+        ].includes(key),
+    ),
+  )
+  if (Object.keys(options).length === 0) return {}
+  return { providerOptions: { openrouter: options } }
+}
+
+function isStringRecord(value: unknown): value is Readonly<Record<string, string>> {
+  return isRecord(value) && Object.values(value).every((item) => typeof item === "string")
+}
+
 function mapXAIOptions(settings: Readonly<Record<string, unknown>>) {
  const options = {
    ...(typeof settings.reasoningEffort === "string" ? { reasoningEffort: settings.reasoningEffort } : {}),
@@ -78,13 +208,3 @@ function mapXAIOptions(settings: Readonly<Record<string, unknown>>) {
  if (Object.keys(options).length === 0) return {}
  return { providerOptions: { xai: options } }
 }
-
-function mapProviderOptions(namespace: string, settings: Readonly<Record<string, unknown>>) {
-  const values = Object.fromEntries(
-    Object.entries(settings).filter(
-      ([key]) => !["apiKey", "authToken", "baseURL", "chunkTimeout", "fetch", "timeout"].includes(key),
-    ),
-  )
-  if (Object.keys(values).length === 0) return {}
-  return { providerOptions: { [namespace]: values } }
-}
@@ -299,8 +299,6 @@ export const locationLayer = Layer.effect(
  }),
 )

-export const defaultLayer = locationLayer
-
 function modelFromLanguage(info: Info, language: LanguageModelV3) {
  const packageName = Provider.packageName(info.package!)
  const projected = mapBodyToProviderOptions(info, packageName)
@@ -332,7 +330,7 @@ function modelFromLanguage(info: Info, language: LanguageModelV3) {
              body: projected.body === undefined ? undefined : { ...projected.body },
              headers: info.headers,
            },
-      limits: { context: info.limit.context, output: info.limit.output },
+      limits: { context: info.limit.context, input: info.limit.input, output: info.limit.output },
      providerOptions,
    },
    body: {
@@ -16,7 +16,7 @@ import { Global } from "@opencode-ai/util/global"
 import { Location } from "./location"
 import { AbsolutePath } from "./schema"
 import { ConfigAgent } from "./config/agent"
-import { ConfigAttachments } from "./config/attachments"
+import { ConfigMedia } from "./config/media"
 import { ConfigCompaction } from "./config/compaction"
 import { ConfigCommand } from "./config/command"
 import { ConfigExperimental } from "./config/experimental"
@@ -85,8 +85,8 @@ export class Info extends Schema.Class<Info>("Config.Info")({
  lsp: ConfigLSP.Info.pipe(Schema.optional).annotate({
    description: "Enable built-in language servers or configure server overrides",
  }),
-  attachments: ConfigAttachments.Info.pipe(Schema.optional).annotate({
-    description: "Attachment processing configuration",
+  media: ConfigMedia.Info.pipe(Schema.optional).annotate({
+    description: "Media processing configuration",
  }),
  tool_output: ConfigToolOutput.Info.pipe(Schema.optional).annotate({
    description: "Tool output truncation thresholds",
@@ -1,15 +1,15 @@
-export * as ConfigAttachments from "./attachments"
+export * as ConfigMedia from "./media"

 import { Schema } from "effect"
 import { PositiveInt } from "../schema"

-export class Image extends Schema.Class<Image>("Config.Attachments.Image")({
+export class Image extends Schema.Class<Image>("Config.Media.Image")({
  auto_resize: Schema.Boolean.pipe(Schema.optional),
  max_width: PositiveInt.pipe(Schema.optional),
  max_height: PositiveInt.pipe(Schema.optional),
  max_base64_bytes: PositiveInt.pipe(Schema.optional),
 }) {}

-export class Info extends Schema.Class<Info>("Config.Attachments")({
+export class Info extends Schema.Class<Info>("Config.Media")({
  image: Image.pipe(Schema.optional),
 }) {}
@@ -1,181 +0,0 @@
-export * as MoveSession from "./move-session"
-
-import { Context, DateTime, Effect, Layer, Schema } from "effect"
-import { makeGlobalNode } from "@opencode-ai/util/effect/app-node"
-import { FSUtil } from "@opencode-ai/util/fs-util"
-import { Git } from "../git"
-import { Global } from "@opencode-ai/util/global"
-import { Project } from "../project"
-import { Session } from "../session"
-import { SessionExecution } from "../session/execution"
-import { SessionSchema } from "../session/schema"
-import { SessionStore } from "../session/store"
-import { AbsolutePath } from "../schema"
-import path from "path"
-
-export const Destination = Schema.Struct({
-  directory: AbsolutePath,
-}).annotate({ identifier: "MoveSession.Destination" })
-export type Destination = typeof Destination.Type
-
-export const Input = Schema.Struct({
-  sessionID: SessionSchema.ID,
-  destination: Destination,
-  moveChanges: Schema.optional(Schema.Boolean),
-}).annotate({ identifier: "MoveSession.Input" })
-export type Input = typeof Input.Type
-
-export class DestinationProjectMismatchError extends Schema.TaggedErrorClass<DestinationProjectMismatchError>()(
-  "MoveSession.DestinationProjectMismatchError",
-  {
-    expected: Project.ID,
-    actual: Project.ID,
-  },
-) {}
-
-export class DestinationNotFoundError extends Schema.TaggedErrorClass<DestinationNotFoundError>()(
-  "MoveSession.DestinationNotFoundError",
-  { directory: AbsolutePath },
-) {}
-
-export class DestinationNotDirectoryError extends Schema.TaggedErrorClass<DestinationNotDirectoryError>()(
-  "MoveSession.DestinationNotDirectoryError",
-  { directory: AbsolutePath },
-) {}
-
-export class ApplyChangesError extends Schema.TaggedErrorClass<ApplyChangesError>()("MoveSession.ApplyChangesError", {
-  message: Schema.String,
-}) {}
-
-export class CaptureChangesError extends Schema.TaggedErrorClass<CaptureChangesError>()(
-  "MoveSession.CaptureChangesError",
-  {
-    message: Schema.String,
-  },
-) {}
-
-export class ResetSourceChangesError extends Schema.TaggedErrorClass<ResetSourceChangesError>()(
-  "MoveSession.ResetSourceChangesError",
-  {
-    directory: AbsolutePath,
-    message: Schema.String,
-    cause: Schema.optional(Schema.Defect()),
-  },
-) {}
-
-export type Error =
-  | Session.NotFoundError
-  | DestinationProjectMismatchError
-  | DestinationNotFoundError
-  | DestinationNotDirectoryError
-  | Session.DestinationNotFoundError
-  | Session.DestinationNotDirectoryError
-  | CaptureChangesError
-  | ApplyChangesError
-  | ResetSourceChangesError
-
-export interface Interface {
-  readonly moveSession: (input: Input) => Effect.Effect<void, Error>
-}
-
-export class Service extends Context.Service<Service, Interface>()("@opencode/ControlPlaneMoveSession") {}
-
-const layer = Layer.effect(
-  Service,
-  Effect.gen(function* () {
-    const git = yield* Git.Service
-    const fs = yield* FSUtil.Service
-    const global = yield* Global.Service
-    const project = yield* Project.Service
-    const sessions = yield* SessionStore.Service
-    const session = yield* Session.Service
-    const execution = yield* SessionExecution.Service
-
-    const moveSession = Effect.fn("MoveSession.moveSession")(function* (input: Input) {
-      const current = yield* sessions.get(input.sessionID)
-      if (!current) return yield* new Session.NotFoundError({ sessionID: input.sessionID })
-      const value = input.destination.directory.trim()
-      const expanded = value === "~" ? global.home : value.startsWith("~/") ? path.join(global.home, value.slice(2)) : value
-      const directory = AbsolutePath.make(path.resolve(current.location.directory, expanded))
-      const destinationInfo = yield* fs.stat(directory).pipe(Effect.catch(() => Effect.succeed(undefined)))
-      if (!destinationInfo) return yield* new DestinationNotFoundError({ directory })
-      if (destinationInfo.type !== "Directory") return yield* new DestinationNotDirectoryError({ directory })
-      if (current.location.directory === directory) return
-
-      const source = yield* project.resolve(current.location.directory)
-      const destination = yield* project.resolve(directory)
-      if (input.moveChanges && current.projectID !== destination.id) {
-        return yield* new DestinationProjectMismatchError({ expected: current.projectID, actual: destination.id })
-      }
-      // A move must not race active execution: a mid-drain relocation would let
-      // the source Location dispatch a request assembled under stale instructions
-      // and history. Serialize like removal does — stop the drain, then move.
-      yield* execution.interrupt(input.sessionID)
-      yield* execution.awaitIdle(input.sessionID)
-
-      const moveChanges = input.moveChanges && source.directory !== destination.directory
-      const sourceRepository = moveChanges ? yield* git.repo.discover(current.location.directory) : undefined
-      if (moveChanges && !sourceRepository)
-        return yield* new CaptureChangesError({ message: "Source is not a Git repository" })
-      const patch = sourceRepository
-        ? yield* git.change
-            .capture({ repository: sourceRepository, path: current.location.directory })
-            .pipe(Effect.mapError((error) => new CaptureChangesError({ message: error.message })))
-        : Git.ChangeSet.make("")
-      if (patch) {
-        const repository = yield* git.repo.discover(directory)
-        if (!repository) return yield* new ApplyChangesError({ message: "Destination is not a Git repository" })
-        yield* git.change
-          .apply({ repository, path: directory, changes: patch })
-          .pipe(Effect.mapError((error) => new ApplyChangesError({ message: error.message })))
-      }
-
-      yield* session.move({
-        sessionID: input.sessionID,
-        directory,
-      })
-
-      if (patch) {
-        const repository = yield* git.repo.discover(current.location.directory)
-        if (!repository)
-          return yield* new ResetSourceChangesError({
-            directory: current.location.directory,
-            message: "Source is not a Git repository",
-          })
-        yield* git.change
-          .discard({
-            repository,
-            path: current.location.directory,
-            index: "preserve",
-            untracked: "remove",
-          })
-          .pipe(
-            Effect.mapError(
-              (error) =>
-                new ResetSourceChangesError({
-                  directory: current.location.directory,
-                  message: error.message,
-                  cause: error.cause,
-                }),
-            ),
-          )
-      }
-    })
-
-    return Service.of({ moveSession })
-  }),
-)
-
-export const node = makeGlobalNode({
-  service: Service,
-  layer,
-  deps: [
-    FSUtil.node,
-    Git.node,
-    Global.node,
-    Project.node,
-    Session.node,
-    SessionStore.node,
-    SessionExecution.node,
-  ],
-})
@@ -1,4 +1,3 @@
-import { createProviderToolFactoryWithOutputSchema } from "@ai-sdk/provider-utils"
 import { z } from "zod/v4"

 export const imageGenerationArgsSchema = z
@@ -24,91 +23,3 @@ export const imageGenerationArgsSchema = z
 export const imageGenerationOutputSchema = z.object({
  result: z.string(),
 })
-
-type ImageGenerationArgs = {
-  /**
-   * Background type for the generated image. Default is 'auto'.
-   */
-  background?: "auto" | "opaque" | "transparent"
-
-  /**
-   * Input fidelity for the generated image. Default is 'low'.
-   */
-  inputFidelity?: "low" | "high"
-
-  /**
-   * Optional mask for inpainting.
-   * Contains image_url (string, optional) and file_id (string, optional).
-   */
-  inputImageMask?: {
-    /**
-     * File ID for the mask image.
-     */
-    fileId?: string
-
-    /**
-     * Base64-encoded mask image.
-     */
-    imageUrl?: string
-  }
-
-  /**
-   * The image generation model to use. Default: gpt-image-1.
-   */
-  model?: string
-
-  /**
-   * Moderation level for the generated image. Default: auto.
-   */
-  moderation?: "auto"
-
-  /**
-   * Compression level for the output image. Default: 100.
-   */
-  outputCompression?: number
-
-  /**
-   * The output format of the generated image. One of png, webp, or jpeg.
-   * Default: png
-   */
-  outputFormat?: "png" | "jpeg" | "webp"
-
-  /**
-   * Number of partial images to generate in streaming mode, from 0 (default value) to 3.
-   */
-  partialImages?: number
-
-  /**
-   * The quality of the generated image.
-   * One of low, medium, high, or auto. Default: auto.
-   */
-  quality?: "auto" | "low" | "medium" | "high"
-
-  /**
-   * The size of the generated image.
-   * One of 1024x1024, 1024x1536, 1536x1024, or auto.
-   * Default: auto.
-   */
-  size?: "auto" | "1024x1024" | "1024x1536" | "1536x1024"
-}
-
-const imageGenerationToolFactory = createProviderToolFactoryWithOutputSchema<
-  {},
-  {
-    /**
-     * The generated image encoded in base64.
-     */
-    result: string
-  },
-  ImageGenerationArgs
->({
-  id: "openai.image_generation",
-  inputSchema: z.object({}),
-  outputSchema: imageGenerationOutputSchema,
-})
-
-export const imageGeneration = (
-  args: ImageGenerationArgs = {}, // default
-) => {
-  return imageGenerationToolFactory(args)
-}
@@ -1,4 +1,3 @@
-import { createProviderToolFactoryWithOutputSchema } from "@ai-sdk/provider-utils"
 import { z } from "zod/v4"

 export const localShellInputSchema = z.object({
@@ -15,50 +14,3 @@ export const localShellInputSchema = z.object({
 export const localShellOutputSchema = z.object({
  output: z.string(),
 })
-
-export const localShell = createProviderToolFactoryWithOutputSchema<
-  {
-    /**
-     * Execute a shell command on the server.
-     */
-    action: {
-      type: "exec"
-
-      /**
-       * The command to run.
-       */
-      command: string[]
-
-      /**
-       * Optional timeout in milliseconds for the command.
-       */
-      timeoutMs?: number
-
-      /**
-       * Optional user to run the command as.
-       */
-      user?: string
-
-      /**
-       * Optional working directory to run the command in.
-       */
-      workingDirectory?: string
-
-      /**
-       * Environment variables to set for the command.
-       */
-      env?: Record<string, string>
-    }
-  },
-  {
-    /**
-     * The output of local shell tool call.
-     */
-    output: string
-  },
-  {}
->({
-  id: "openai.local_shell",
-  inputSchema: localShellInputSchema,
-  outputSchema: localShellOutputSchema,
-})
@@ -1,4 +1,3 @@
-import { createProviderToolFactory } from "@ai-sdk/provider-utils"
 import { z } from "zod/v4"

 // Args validation schema
@@ -39,65 +38,3 @@ export const webSearchPreviewArgsSchema = z.object({
    })
    .optional(),
 })
-
-export const webSearchPreview = createProviderToolFactory<
-  {
-    // Web search doesn't take input parameters - it's controlled by the prompt
-  },
-  {
-    /**
-     * Search context size to use for the web search.
-     * - high: Most comprehensive context, highest cost, slower response
-     * - medium: Balanced context, cost, and latency (default)
-     * - low: Least context, lowest cost, fastest response
-     */
-    searchContextSize?: "low" | "medium" | "high"
-
-    /**
-     * User location information to provide geographically relevant search results.
-     */
-    userLocation?: {
-      /**
-       * Type of location (always 'approximate')
-       */
-      type: "approximate"
-      /**
-       * Two-letter ISO country code (e.g., 'US', 'GB')
-       */
-      country?: string
-      /**
-       * City name (free text, e.g., 'Minneapolis')
-       */
-      city?: string
-      /**
-       * Region name (free text, e.g., 'Minnesota')
-       */
-      region?: string
-      /**
-       * IANA timezone (e.g., 'America/Chicago')
-       */
-      timezone?: string
-    }
-  }
->({
-  id: "openai.web_search_preview",
-  inputSchema: z.object({
-    action: z
-      .discriminatedUnion("type", [
-        z.object({
-          type: z.literal("search"),
-          query: z.string().nullish(),
-        }),
-        z.object({
-          type: z.literal("open_page"),
-          url: z.string(),
-        }),
-        z.object({
-          type: z.literal("find"),
-          url: z.string(),
-          pattern: z.string(),
-        }),
-      ])
-      .nullish(),
-  }),
-})
@@ -1,4 +1,3 @@
-import { createProviderToolFactory } from "@ai-sdk/provider-utils"
 import { z } from "zod/v4"

 export const webSearchArgsSchema = z.object({
@@ -20,83 +19,3 @@ export const webSearchArgsSchema = z.object({
    })
    .optional(),
 })
-
-export const webSearchToolFactory = createProviderToolFactory<
-  {
-    // Web search doesn't take input parameters - it's controlled by the prompt
-  },
-  {
-    /**
-     * Filters for the search.
-     */
-    filters?: {
-      /**
-       * Allowed domains for the search.
-       * If not provided, all domains are allowed.
-       * Subdomains of the provided domains are allowed as well.
-       */
-      allowedDomains?: string[]
-    }
-
-    /**
-     * Search context size to use for the web search.
-     * - high: Most comprehensive context, highest cost, slower response
-     * - medium: Balanced context, cost, and latency (default)
-     * - low: Least context, lowest cost, fastest response
-     */
-    searchContextSize?: "low" | "medium" | "high"
-
-    /**
-     * User location information to provide geographically relevant search results.
-     */
-    userLocation?: {
-      /**
-       * Type of location (always 'approximate')
-       */
-      type: "approximate"
-      /**
-       * Two-letter ISO country code (e.g., 'US', 'GB')
-       */
-      country?: string
-      /**
-       * City name (free text, e.g., 'Minneapolis')
-       */
-      city?: string
-      /**
-       * Region name (free text, e.g., 'Minnesota')
-       */
-      region?: string
-      /**
-       * IANA timezone (e.g., 'America/Chicago')
-       */
-      timezone?: string
-    }
-  }
->({
-  id: "openai.web_search",
-  inputSchema: z.object({
-    action: z
-      .discriminatedUnion("type", [
-        z.object({
-          type: z.literal("search"),
-          query: z.string().nullish(),
-        }),
-        z.object({
-          type: z.literal("open_page"),
-          url: z.string(),
-        }),
-        z.object({
-          type: z.literal("find"),
-          url: z.string(),
-          pattern: z.string(),
-        }),
-      ])
-      .nullish(),
-  }),
-})
-
-export const webSearch = (
-  args: Parameters<typeof webSearchToolFactory>[0] = {}, // default
-) => {
-  return webSearchToolFactory(args)
-}
@@ -61,7 +61,7 @@ const layer = Layer.effect(
      const image = Object.assign(
        {},
        ...(yield* config.entries()).flatMap((entry) =>
-          entry.type === "document" && entry.info.attachments?.image ? [entry.info.attachments.image] : [],
+          entry.type === "document" && entry.info.media?.image ? [entry.info.media.image] : [],
        ),
      )
      const normalize = yield* loadAdapter
@@ -148,6 +148,3 @@ export function buildLocationServiceMap(
    ),
  )
 }
-
-// This is temporary for backwards compatibility
-export const locationServiceMapLayer = buildLocationServiceMap()
@@ -81,7 +81,7 @@ const withDefaults = (model: Info, route: AnyRoute) =>
    headers: providerHeaders(model),
    providerOptions: providerOptions(model),
    http: model.body === undefined ? undefined : { body: model.body },
-    limits: { context: model.limit.context, output: model.limit.output },
+    limits: { context: model.limit.context, input: model.limit.input, output: model.limit.output },
  })

 const providerHeaders = (model: Info) => {
@@ -96,9 +96,7 @@ const providerHeaders = (model: Info) => {
  return Provider.mergeHeaders(generated.size === 0 ? undefined : Object.fromEntries(generated), model.headers)
 }

-const providerOptions = (
-  model: Info,
-): { readonly [key: string]: { readonly [key: string]: unknown } } | undefined => {
+const providerOptions = (model: Info): { readonly [key: string]: { readonly [key: string]: unknown } } | undefined => {
  if (!Provider.isAISDK(model.package) || model.settings === undefined) return undefined
  const { apiKey: _, baseURL: _baseURL, ...settings } = model.settings
  if (Object.keys(settings).length === 0) return undefined
@@ -177,15 +175,22 @@ export const fromCatalogModel = (
        .model({ id: resolved.modelID ?? resolved.id, compatibility: resolved.compatibility }),
    )
  }
-  const configured = { ...resolved.settings, ...credential?.metadata }
-  const mapping = Provider.isAISDK(resolved.package) ? AISDKNative.map(packageName, configured) : undefined
+  const configured = {
+    ...resolved.settings,
+    ...credential?.metadata,
+    ...(packageName === "@ai-sdk/amazon-bedrock/mantle"
+      ? nativeCredentialSettings("@opencode-ai/ai/providers/amazon-bedrock/mantle", credential)
+      : {}),
+  }
+  const mapping = Provider.isAISDK(resolved.package)
+    ? AISDKNative.map(packageName, configured, resolved.modelID ?? resolved.id)
+    : undefined
  const native = mapping?.package ?? resolved.package
  if (Provider.isAISDK(resolved.package) && !mapping) {
    if (!dependencies?.loadAISDK) return Effect.fail(unsupported(resolved))
    const runtime = produce(resolved, (draft) => {
      draft.settings = Provider.mergeOverlay(draft.settings, {
-        ...(credential?.type === "key" ? { apiKey: credential.key } : {}),
-        ...(credential?.type === "oauth" ? { apiKey: credential.access } : {}),
+        ...nativeCredentialSettings(resolved.package ?? "", credential),
        ...credential?.metadata,
      })
    })
@@ -202,9 +207,9 @@ export const fromCatalogModel = (
    const settings = {
      ...(credential ? withoutNativeAuthSettings(mapped) : mapped),
      ...nativeCredentialSettings(specifier, credential),
-      headers: resolved.headers,
-      body: resolved.body,
-      limits: { context: resolved.limit.context, output: resolved.limit.output },
+      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: () => {
@@ -266,10 +271,7 @@ export const layer = Layer.effect(
    const integrations = yield* Integration.Service
    const npm = yield* Npm.Service
    const aisdk = yield* AISDK.Service
-    const load = Effect.fn("ModelResolver.resolveModel")(function* (
-      selected: Info,
-      variant?: VariantID,
-    ) {
+    const load = Effect.fn("ModelResolver.resolveModel")(function* (selected: Info, variant?: VariantID) {
      const provider = yield* catalog.provider.get(selected.providerID)
      const connection = yield* integrations.connection.active(
        provider?.integrationID ?? Integration.ID.make(selected.providerID),
@@ -341,6 +341,7 @@ export const make = Effect.fn("PluginHost.make")(function* (plugin: import("../p
      create: (input) =>
        runtime.session.create({
          id: input?.id,
+          title: input?.title,
          agent: input?.agent,
          model: input?.model,
          location:
@@ -350,8 +351,10 @@ export const make = Effect.fn("PluginHost.make")(function* (plugin: import("../p
      prompt: runtime.session.prompt,
      generate: (input) => runtime.session.generate(input).pipe(Effect.map((text) => ({ text }))),
      command: runtime.session.command,
+      rename: runtime.session.rename,
      synthetic: runtime.session.synthetic,
      interrupt: (input) => runtime.session.interrupt(input.sessionID),
+      wait: (input) => runtime.session.wait(input.sessionID),
    },
  } satisfies Plugin.Context
 })
@@ -1,94 +0,0 @@
-export * as LayerMapExample from "./layer-map.example"
-
-import { Context, Effect, Layer, LayerMap } from "effect"
-import { Npm } from "@opencode-ai/util/npm"
-
-/**
- * Tutorial: split global services from context-specific services.
- *
- * Use this pattern when part of the app should be constructed once at the app edge,
- * while another part should be cached per request/project/workspace key.
- *
- * In this example:
- * - Npm.Service is the global service. It is not keyed by request context and should
- *   be provided once by the application runtime.
- * - ConfigService is context-specific. It is built from a RequestContext key and is
- *   cached by LayerMap for that key.
- * - ConfigServiceMap.layer owns the cache. Provide it once globally, then each
- *   request can provide ConfigServiceMap.get(context) to select the right instance.
- *
- * Lifetime model:
- * - ConfigServiceMap.layer has the app/global lifetime and depends on Npm.Service.
- * - ConfigServiceMap.get(context) has the request/context lifetime and provides
- *   ConfigService for exactly that context key.
- * - The cached ConfigService entry stays alive while something is using it. Once idle,
- *   it remains cached for idleTimeToLive, then its scope is finalized.
- * - invalidate(context) removes the cache entry for future lookups. Active users keep
- *   running on the old instance; the next lookup can create a fresh instance.
- *
- * Key model:
- * - Keys can be strings, structs, classes, arrays, etc.
- * - Prefer primitive or immutable keys. Effect uses Hash / Equal semantics for cache
- *   lookup, so mutating an object after it has been used as a key is a bug.
- */
-
-export type RequestContext = {
-  readonly directory: string
-  readonly workspace: string
-}
-
-export class RequestContextRef extends Context.Service<RequestContextRef, RequestContext>()(
-  "@opencode/example/RequestContextRef",
-) {}
-
-export interface ConfigServiceShape {
-  readonly directory: string
-  readonly workspace: string
-  readonly nextUse: () => Effect.Effect<number>
-  readonly which: Npm.Interface["which"]
-}
-
-export class ConfigService extends Context.Service<ConfigService, ConfigServiceShape>()(
-  "@opencode/example/ConfigService",
-) {}
-
-const configServiceLayer = Layer.effect(
-  ConfigService,
-  Effect.gen(function* () {
-    const context = yield* RequestContextRef
-    const npm = yield* Npm.Service
-
-    let useCount = 0
-
-    return ConfigService.of({
-      directory: context.directory,
-      workspace: context.workspace,
-      nextUse: () => Effect.succeed(++useCount),
-      which: npm.which,
-    })
-  }),
-)
-
-export class ConfigServiceMap extends LayerMap.Service<ConfigServiceMap>()("@opencode/example/ConfigServiceMap", {
-  lookup: (context: RequestContext) =>
-    configServiceLayer.pipe(Layer.provide(Layer.succeed(RequestContextRef, RequestContextRef.of(context)))),
-  idleTimeToLive: "5 minutes",
-}) {}
-
-export const appLayer = ConfigServiceMap.layer
-
-export const readConfig = Effect.fn("LayerMapExample.readConfig")(function* () {
-  const config = yield* ConfigService
-
-  return {
-    directory: config.directory,
-    workspace: config.workspace,
-    useCount: yield* config.nextUse(),
-  }
-})
-
-export const handleRequest = Effect.fn("LayerMapExample.handleRequest")(function* (context: RequestContext) {
-  return yield* readConfig().pipe(Effect.provide(ConfigServiceMap.get(context)))
-})
-
-export const invalidateContext = (context: RequestContext) => ConfigServiceMap.invalidate(context)
@@ -220,12 +220,8 @@ export const OpenAIPlugin = define({
            return
          }
          draft.cost = []
-          if (draft.id.includes("gpt-5.5")) {
-            draft.limit = { context: 400_000, input: 272_000, output: 128_000 }
-          }
-          if (draft.id.includes("gpt-5.6")) {
-            draft.limit = { context: 500_000, input: 372_000, output: 128_000 }
-          }
+          // Match Codex CLI so context consumption and subscription usage stay consistent between clients.
+          draft.limit = { ...draft.limit, context: 272_000, input: 272_000 }
        })
      }
    })
@@ -11,7 +11,17 @@ import { Session } from "../session"
 export interface Interface {
  readonly session: Pick<
    Session.Interface,
-    "get" | "create" | "messages" | "prompt" | "generate" | "command" | "resume" | "interrupt" | "synthetic"
+    | "get"
+    | "create"
+    | "messages"
+    | "prompt"
+    | "generate"
+    | "command"
+    | "rename"
+    | "resume"
+    | "interrupt"
+    | "synthetic"
+    | "wait"
  >
  readonly job: Pick<Job.Interface, "start" | "wait" | "block" | "background" | "cancel">
  readonly location: {
@@ -52,9 +62,11 @@ export const layerWithCell = (cell: Cell) =>
        prompt: (input) => require(cell, (runtime) => runtime.session.prompt(input)),
        generate: (input) => require(cell, (runtime) => runtime.session.generate(input)),
        command: (input) => require(cell, (runtime) => runtime.session.command(input)),
+        rename: (input) => require(cell, (runtime) => runtime.session.rename(input)),
        resume: (sessionID) => require(cell, (runtime) => runtime.session.resume(sessionID)),
        interrupt: (sessionID) => require(cell, (runtime) => runtime.session.interrupt(sessionID)),
        synthetic: (input) => require(cell, (runtime) => runtime.session.synthetic(input)),
+        wait: (sessionID) => require(cell, (runtime) => runtime.session.wait(sessionID)),
      },
      job: {
        start: (input) => require(cell, (runtime) => runtime.job.start(input)),
@@ -108,7 +120,6 @@ export const providerLayerWithCell = (cell: Cell) =>
  )

 export const layer = layerWithCell(defaultCell)
-export const providerLayer = providerLayerWithCell(defaultCell)

 export const node = makeGlobalNode({ service: Service, layer, deps: [] })

@@ -34,6 +34,15 @@ export type ProviderPackage = ProviderPackageDefinition
 const packages = new Map<string, Promise<unknown>>()
 const builtins = new Map<string, () => Promise<unknown>>([
  ["@opencode-ai/ai/providers/amazon-bedrock", () => import("@opencode-ai/ai/providers/amazon-bedrock")],
+  ["@opencode-ai/ai/providers/amazon-bedrock/mantle", () => import("@opencode-ai/ai/providers/amazon-bedrock/mantle")],
+  [
+    "@opencode-ai/ai/providers/amazon-bedrock/mantle/chat",
+    () => import("@opencode-ai/ai/providers/amazon-bedrock/mantle/chat"),
+  ],
+  [
+    "@opencode-ai/ai/providers/amazon-bedrock/mantle/responses",
+    () => import("@opencode-ai/ai/providers/amazon-bedrock/mantle/responses"),
+  ],
  ["@opencode-ai/ai/providers/anthropic", () => import("@opencode-ai/ai/providers/anthropic")],
  ["@opencode-ai/ai/providers/azure", () => import("@opencode-ai/ai/providers/azure")],
  ["@opencode-ai/ai/providers/azure/chat", () => import("@opencode-ai/ai/providers/azure/chat")],
@@ -938,7 +938,7 @@ const materializeAttachment = Effect.fn("Session.materializeAttachment")(functio
      : resolved.bytes
  const normalized = yield* normalizeImageAttachment(
    input,
-    Base64.make(Buffer.from(content).toString("base64")),
+    Buffer.from(content).toString("base64"),
    mime,
    image,
  )
@@ -954,11 +954,11 @@ const materializeAttachment = Effect.fn("Session.materializeAttachment")(functio

 const normalizeImageAttachment = Effect.fn("Session.normalizeImageAttachment")(function* (
  input: PromptInput.FileAttachment,
-  data: Base64,
+  data: string,
  mime: string,
  image: Effect.Effect<Image.Interface>,
 ) {
-  if (!mime.startsWith("image/")) return { data, mime }
+  if (!mime.startsWith("image/")) return { data: Base64.make(data), mime }
  const service = yield* image
  const label = input.name ?? (input.uri.startsWith("data:") ? "inline attachment" : input.uri)
  const content = { uri: label, content: data, encoding: "base64" as const, mime }
@@ -152,14 +152,11 @@ const settings = (documents: readonly Config.Entry[]) => {
  const configured = documents
    .filter((entry): entry is Config.Document => entry.type === "document")
    .flatMap((entry) => (entry.info.compaction ? [entry.info.compaction] : []))
-  return configured.reduce<Settings>(
-    (result, current) => ({
-      auto: current.auto ?? result.auto,
-      buffer: current.buffer ?? result.buffer,
-      tokens: current.keep?.tokens ?? result.tokens,
-    }),
-    { auto: true, buffer: DEFAULT_BUFFER, tokens: DEFAULT_KEEP_TOKENS },
-  )
+  return {
+    auto: configured.findLast((value) => value.auto !== undefined)?.auto ?? true,
+    buffer: configured.findLast((value) => value.buffer !== undefined)?.buffer ?? DEFAULT_BUFFER,
+    tokens: configured.findLast((value) => value.keep?.tokens !== undefined)?.keep?.tokens ?? DEFAULT_KEEP_TOKENS,
+  }
 }

 const select = (
@@ -350,11 +347,16 @@ const make = (dependencies: Dependencies) => {
        message.type === "assistant" && message.tokens !== undefined,
    )
    if (!last) return false
-    const output = Math.min(input.model.route.defaults.limits?.output ?? 0, OUTPUT_TOKEN_MAX)
+    const limits = input.model.route.defaults.limits
+    const output = Math.min(limits?.output ?? 0, OUTPUT_TOKEN_MAX)
+    const promptCeiling = Math.min(
+      limits?.input === undefined ? Number.POSITIVE_INFINITY : limits.input - config.buffer,
+      context - Math.max(output, config.buffer),
+    )
    const used =
      last.tokens.input + last.tokens.output + last.tokens.reasoning + last.tokens.cache.read + last.tokens.cache.write
    if (used <= 0) return false
-    return used >= context - (output || config.buffer)
+    return used >= promptCeiling
  }
  const compactManual = Effect.fn("SessionCompaction.compactManual")(function* (input: ManualInput) {
    const content = planContent(input.messages, config.tokens)
@@ -71,7 +71,7 @@ export const layer = Layer.effect(
          LLM.request({
            model: model.model,
            http: { headers: SessionModelHeaders.make(selection.session, app) },
-            providerOptions: { openai: { promptCacheKey } },
+            providerOptions: { [providerMetadataKey]: { promptCacheKey } },
            system: contextEvent.system,
            messages: contextEvent.messages,
            tools: hookedTools,
@@ -19,6 +19,11 @@ import { MAX_STEPS_PROMPT } from "./runner/max-steps"
 import PROMPT_DEFAULT from "./runner/prompt/base.txt"
 import { toLLMMessages } from "./runner/to-llm-message"

+const IMAGE_BYTES_TRIGGER = 25 * 1024 * 1024 // 25 MiB
+const IMAGE_BYTES_TARGET = 15 * 1024 * 1024 // 15 MiB
+const IMAGE_REMOVED =
+  "[This image was removed to reduce the request size and is no longer visible. Do not make claims about its contents from memory. If needed, retrieve it again with an available tool or ask the user to attach it again.]"
+
 /** Failures a prepared execution can surface: infrastructure errors plus user declines resurfaced from the defect tunnel. */
 export type ExecuteError = Tool.Error | Permission.DeclinedError | QuestionTool.CancelledError

@@ -94,6 +99,56 @@ export const unsupportedParts = (messages: LLMRequest["messages"], capabilities:
    }),
  )

+export const boundImages = (messages: LLMRequest["messages"]) => {
+  const isImage = (mime: string) => mime.toLowerCase().startsWith("image/")
+  const size = (data: string | Uint8Array) =>
+    typeof data === "string" ? Buffer.byteLength(data) : Math.ceil(data.byteLength / 3) * 4
+  const imageBytes = messages.reduce(
+    (total, message) =>
+      total +
+      message.content.reduce((sum, part) => {
+        if (part.type === "media" && isImage(part.mediaType)) return sum + size(part.data)
+        if (part.type !== "tool-result" || part.result.type !== "content") return sum
+        return (
+          sum +
+          part.result.value.reduce(
+            (bytes: number, item: Content) =>
+              bytes + (item.type === "file" && isImage(item.mime) ? Buffer.byteLength(item.uri) : 0),
+            0,
+          )
+        )
+      }, 0),
+    0,
+  )
+  if (imageBytes <= IMAGE_BYTES_TRIGGER) return messages
+
+  let removed = 0
+  return messages.map((message) =>
+    Message.make({
+      ...message,
+      content: message.content.map((part) => {
+        if (part.type === "media" && isImage(part.mediaType) && imageBytes - removed > IMAGE_BYTES_TARGET) {
+          removed += size(part.data)
+          return Message.text(IMAGE_REMOVED)
+        }
+        if (part.type !== "tool-result" || part.result.type !== "content") return part
+        return {
+          ...part,
+          result: {
+            ...part.result,
+            value: part.result.value.map((item: Content) => {
+              if (item.type !== "file" || !isImage(item.mime) || imageBytes - removed <= IMAGE_BYTES_TARGET)
+                return item
+              removed += Buffer.byteLength(item.uri)
+              return { type: "text" as const, text: IMAGE_REMOVED }
+            }),
+          },
+        }
+      }),
+    }),
+  )
+}
+
 /**
 * Builds an outbound model request and captures the tool-call capability that
 * must remain paired with it. It does not execute the request or mutate
@@ -158,9 +213,9 @@ export const layer = Layer.effect(
        http: {
          headers: SessionModelHeaders.make(session, app),
        },
-        providerOptions: { openai: { promptCacheKey } },
+        providerOptions: { [providerMetadataKey]: { promptCacheKey } },
        system: contextEvent.system,
-        messages: unsupportedParts(contextEvent.messages, resolved.capabilities),
+        messages: boundImages(unsupportedParts(contextEvent.messages, resolved.capabilities)),
        tools: hookedTools,
        toolChoice: stepLimitReached ? "none" : undefined,
      })
@@ -6,6 +6,7 @@ import { Agent } from "../agent"
 import { Database } from "../database/database"
 import { Bus } from "../bus"
 import { makeLocationNode } from "@opencode-ai/util/effect/app-node"
+import { isExactRootFallback } from "@opencode-ai/util/session-title-fallback"
 import { App } from "../app"
 import { llmClient } from "../effect/app-node-platform"
 import { SessionEvent } from "./event"
@@ -39,7 +40,10 @@ export class Service extends Context.Service<Service, Interface>()("@opencode/Se

 const truncate = (value: string) => (value.length <= MAX_LENGTH ? value : `${value.slice(0, MAX_LENGTH - 3)}...`)
 const isUntitled = (session: SessionSchema.Info) =>
-  session.title === undefined || session.title === `New session - ${DateTime.formatIso(session.time.created)}`
+  isExactRootFallback({
+    title: session.title,
+    time: { created: DateTime.toEpochMillis(session.time.created) },
+  })

 const make = (dependencies: Dependencies) => {
  const generateForFirstPrompt = Effect.fn("SessionTitle.generateForFirstPrompt")(function* (
@@ -11,25 +11,25 @@ export function toSessionError(cause: unknown): SessionError.Error {
  if (cause instanceof LLMError) {
    switch (cause.reason._tag) {
      case "RateLimit":
-        return { type: "provider.rate-limit", message: cause.reason.message }
+        return providerError("provider.rate-limit", cause.reason)
      case "Authentication":
-        return { type: "provider.auth", message: cause.reason.message }
+        return providerError("provider.auth", cause.reason)
      case "QuotaExceeded":
-        return { type: "provider.quota", message: cause.reason.message }
+        return providerError("provider.quota", cause.reason)
      case "ContentPolicy":
-        return { type: "provider.content-filter", message: cause.reason.message }
+        return providerError("provider.content-filter", cause.reason)
      case "Transport":
-        return { type: "provider.transport", message: cause.reason.message }
+        return providerError("provider.transport", cause.reason)
      case "ProviderInternal":
-        return { type: "provider.internal", message: cause.reason.message }
+        return providerError("provider.internal", cause.reason)
      case "InvalidProviderOutput":
-        return { type: "provider.invalid-output", message: cause.reason.message }
+        return providerError("provider.invalid-output", cause.reason)
      case "InvalidRequest":
-        return { type: "provider.invalid-request", message: cause.reason.message }
+        return providerError("provider.invalid-request", cause.reason)
      case "NoRoute":
-        return { type: "provider.no-route", message: cause.reason.message }
+        return providerError("provider.no-route", cause.reason)
      case "UnknownProvider":
-        return { type: "provider.unknown", message: cause.reason.message }
+        return providerError("provider.unknown", cause.reason)
      default: {
        const exhaustive: never = cause.reason
        return exhaustive
@@ -58,3 +58,9 @@ export function toSessionError(cause: unknown): SessionError.Error {
  if (cause instanceof Integration.AuthorizationError) return { type: "provider.auth", message: cause.message }
  return { type: "unknown", message: cause instanceof Error ? cause.message : String(cause) }
 }
+
+function providerError(type: string, reason: LLMError["reason"]): SessionError.Error {
+  const status =
+    ("http" in reason ? reason.http?.response?.status : undefined) ?? ("status" in reason ? reason.status : undefined)
+  return { type, message: reason.message, ...(status === undefined ? {} : { status }) }
+}
@@ -1,10 +0,0 @@
-import { createRequire } from "node:module"
-import path from "node:path"
-
-export namespace Module {
-  export function resolve(id: string, dir: string) {
-    try {
-      return createRequire(path.join(dir, "package.json")).resolve(id)
-    } catch {}
-  }
-}
@@ -12,12 +12,6 @@ export function getDirectory(path: string | undefined) {
  return parts.slice(0, parts.length - 1).join("/") + "/"
 }

-export function getFileExtension(path: string | undefined) {
-  if (!path) return ""
-  const parts = path.split(".")
-  return parts[parts.length - 1]
-}
-
 export function getFilenameTruncated(path: string | undefined, maxLength: number = 20) {
  const filename = getFilename(path)
  if (filename.length <= maxLength) return filename
@@ -1,16 +0,0 @@
-export * as ConfigConsoleStateV1 from "./console-state"
-
-import { Schema } from "effect"
-import { NonNegativeInt } from "../../schema"
-
-export class ConsoleState extends Schema.Class<ConsoleState>("ConsoleState")({
-  consoleManagedProviders: Schema.mutable(Schema.Array(Schema.String)),
-  activeOrgName: Schema.optional(Schema.String),
-  switchableOrgCount: NonNegativeInt,
-}) {}
-
-export const emptyConsoleState: ConsoleState = ConsoleState.make({
-  consoleManagedProviders: [],
-  activeOrgName: undefined,
-  switchableOrgCount: 0,
-})
@@ -63,7 +63,7 @@ export function migrate(info: typeof ConfigV1.Info.Type) {
    watcher: info.watcher,
    formatter: info.formatter,
    lsp: info.lsp,
-    attachments: info.attachment,
+    media: info.attachment,
    tool_output: info.tool_output,
    mcp: mcp(info),
    compaction: info.compaction && {
@@ -2,6 +2,123 @@ import { describe, expect, test } from "bun:test"
 import { AISDKNative } from "@opencode-ai/core/aisdk-native"

 describe("AISDKNative", () => {
+  test("maps Bedrock Mantle models to their supported native APIs", () => {
+    const settings = {
+      bearerToken: "token",
+      region: "us-west-2",
+      baseURL: "https://mantle.test/v1",
+      reasoningEffort: "high",
+      reasoningSummary: "auto",
+      include: ["reasoning.encrypted_content"],
+    }
+
+    expect(AISDKNative.map("@ai-sdk/amazon-bedrock/mantle", settings, "openai.gpt-oss-120b")).toEqual({
+      package: "@opencode-ai/ai/providers/amazon-bedrock/mantle/responses",
+      settings: {
+        apiKey: "token",
+        baseURL: "https://mantle.test/v1",
+        region: "us-west-2",
+        providerOptions: {
+          openai: {
+            reasoningEffort: "high",
+            reasoningSummary: "auto",
+            include: ["reasoning.encrypted_content"],
+          },
+        },
+      },
+    })
+    expect(AISDKNative.map("@ai-sdk/amazon-bedrock/mantle", settings, "openai.gpt-oss-safeguard-20b")?.package).toBe(
+      "@opencode-ai/ai/providers/amazon-bedrock/mantle/chat",
+    )
+  })
+
+  test("maps static Bedrock Mantle credentials without leaking connection options", () => {
+    expect(
+      AISDKNative.map(
+        "@ai-sdk/amazon-bedrock/mantle",
+        {
+          accessKeyId: "key",
+          secretAccessKey: "secret",
+          sessionToken: "session",
+          region: "eu-west-1",
+          profile: "ignored",
+          credentialProvider: "ignored",
+          fetch: "ignored",
+          store: false,
+        },
+        "openai.gpt-oss-120b",
+      ),
+    ).toEqual({
+      package: "@opencode-ai/ai/providers/amazon-bedrock/mantle/responses",
+      settings: {
+        credentials: {
+          accessKeyId: "key",
+          secretAccessKey: "secret",
+          sessionToken: "session",
+          region: "eu-west-1",
+        },
+        region: "eu-west-1",
+        providerOptions: { openai: { store: false } },
+      },
+    })
+  })
+
+  test("keeps Bedrock Mantle on the AI SDK when native static auth is unavailable", () => {
+    expect(
+      AISDKNative.map(
+        "@ai-sdk/amazon-bedrock/mantle",
+        { region: "us-east-1", profile: "production" },
+        "openai.gpt-oss-120b",
+      ),
+    ).toBeUndefined()
+  })
+
+  test("maps the legacy Bedrock endpoint override", () => {
+    expect(
+      AISDKNative.map(
+        "@ai-sdk/amazon-bedrock/mantle",
+        { bearerToken: "token", endpoint: "https://mantle.private/v1", region: "us-east-1" },
+        "openai.gpt-oss-120b",
+      ),
+    ).toMatchObject({ settings: { baseURL: "https://mantle.private/v1" } })
+  })
+
+  test("maps OpenRouter settings to native destinations", () => {
+    expect(
+      AISDKNative.map("@openrouter/ai-sdk-provider", {
+        appName: "OpenCode",
+        appUrl: "https://opencode.ai",
+        headers: { "x-openrouter-title": "Configured", "x-provider-api-keys": "Configured BYOK" },
+        api_keys: { anthropic: "provider-key" },
+        extraBody: { transforms: ["middle-out"] },
+        models: ["anthropic/claude-sonnet-4.6"],
+        provider: { only: ["anthropic"], require_parameters: true },
+        reasoning: { effort: "high" },
+        promptCacheKey: "session_123",
+        future_option: { enabled: true },
+      }),
+    ).toEqual({
+      package: "@opencode-ai/ai/providers/openrouter",
+      settings: {
+        providerOptions: {
+          openrouter: {
+            models: ["anthropic/claude-sonnet-4.6"],
+            provider: { only: ["anthropic"], require_parameters: true },
+            reasoning: { effort: "high" },
+            promptCacheKey: "session_123",
+            future_option: { enabled: true },
+          },
+        },
+      },
+      headers: {
+        "x-openrouter-title": "Configured",
+        "HTTP-Referer": "https://opencode.ai",
+        "x-provider-api-keys": "Configured BYOK",
+      },
+      body: { transforms: ["middle-out"] },
+    })
+  })
+
  test("maps every Google thinking setting", () => {
    expect(
      AISDKNative.map("@ai-sdk/google", {
@@ -803,7 +803,7 @@ describe("Config", () => {
                  custom: { command: ["custom-fmt", "$FILE"], extensions: [".foo"] },
                },
                lsp: { typescript: { disabled: true }, custom: { command: ["custom-lsp"], extensions: [".foo"] } },
-                attachments: {
+                media: {
                  image: { auto_resize: false, max_width: 1200, max_height: 900, max_base64_bytes: 1048576 },
                },
                tool_output: { max_lines: 1000, max_bytes: 32768 },
@@ -890,7 +890,7 @@ describe("Config", () => {
              typescript: { disabled: true },
              custom: { command: ["custom-lsp"], extensions: [".foo"] },
            })
-            expect(documents[0]?.info.attachments).toEqual({
+            expect(documents[0]?.info.media).toEqual({
              image: { auto_resize: false, max_width: 1200, max_height: 900, max_base64_bytes: 1048576 },
            })
            expect(documents[0]?.info.tool_output).toEqual({ max_lines: 1000, max_bytes: 32768 })
@@ -1097,7 +1097,7 @@ describe("Config", () => {
            expect(documents[0]?.info.references).toEqual({
              docs: { path: "../docs", description: "Use for product documentation", hidden: true },
            })
-            expect(documents[0]?.info.attachments).toEqual({ image: { auto_resize: false, max_width: 1200 } })
+            expect(documents[0]?.info.media).toEqual({ image: { auto_resize: false, max_width: 1200 } })
            expect(documents[0]?.info.providers?.custom).toMatchObject({
              settings: { apiKey: "secret" },
              models: {
@@ -1,5 +1,6 @@
 import { describe, expect } from "bun:test"
 import { LLM, Model } from "@opencode-ai/ai"
+import { OpenAIChat } from "@opencode-ai/ai/protocols"
 import { compileRequest } from "@opencode-ai/ai/route/client"
 import { Effect } from "effect"
 import { Headers } from "effect/unstable/http"
@@ -17,6 +18,7 @@ interface ModelOptions {
  readonly headers?: Info["headers"]
  readonly body?: Info["body"]
  readonly variants?: Info["variants"]
+  readonly limit?: Info["limit"]
 }

 const model = (packageName: string | undefined, options: ModelOptions = {}) =>
@@ -36,14 +38,44 @@ const model = (packageName: string | undefined, options: ModelOptions = {}) =>
    cost: [],
    status: "active",
    enabled: true,
-    limit: { context: 100, output: 20 },
+    limit: options.limit ?? { context: 100, output: 20 },
  })

 describe("ModelResolver", () => {
+  it.effect("maps Bedrock Mantle models to native Responses and safeguards to Chat", () =>
+    Effect.gen(function* () {
+      const credential = Credential.Key.make({ type: "key", key: "secret" })
+      const responses = yield* ModelResolver.fromCatalogModel(
+        model(Provider.aisdk("@ai-sdk/amazon-bedrock/mantle"), {
+          modelID: "openai.gpt-oss-120b",
+          settings: { region: "us-east-2" },
+        }),
+        credential,
+      )
+      const chat = yield* ModelResolver.fromCatalogModel(
+        model(Provider.aisdk("@ai-sdk/amazon-bedrock/mantle"), {
+          modelID: "openai.gpt-oss-safeguard-20b",
+          settings: { region: "us-east-2" },
+        }),
+        credential,
+      )
+
+      expect(responses.route).toMatchObject({
+        id: "bedrock-mantle-responses",
+        endpoint: { baseURL: "https://bedrock-mantle.us-east-2.api.aws/v1" },
+      })
+      expect(chat.route).toMatchObject({
+        id: "bedrock-mantle-chat",
+        endpoint: { baseURL: "https://bedrock-mantle.us-east-2.api.aws/v1" },
+      })
+    }),
+  )
+
  it.effect("uses the API modelID instead of the catalog ID for native OpenAI routes", () =>
    Effect.gen(function* () {
      const catalog = model(Provider.aisdk("@ai-sdk/openai"), {
        settings: { baseURL: "https://openai.example/v1" },
+        limit: { context: 100, input: 80, output: 20 },
      })
      const resolved = yield* ModelResolver.fromCatalogModel(catalog)

@@ -55,7 +87,7 @@ describe("ModelResolver", () => {
        endpoint: { baseURL: "https://openai.example/v1" },
        defaults: {
          headers: { "x-test": "header" },
-          limits: { context: 100, output: 20 },
+          limits: { context: 100, input: 80, output: 20 },
          http: { body: { custom_extension: { enabled: true } } },
        },
      })
@@ -544,6 +576,37 @@ describe("ModelResolver", () => {
    }),
  )

+  it.effect("merges mapped OpenRouter headers and body with catalog overlays", () =>
+    ModelResolver.fromCatalogModel(
+      model(Provider.aisdk("@openrouter/ai-sdk-provider"), {
+        settings: {
+          appName: "OpenCode",
+          appUrl: "https://opencode.ai",
+          extraBody: { transforms: ["middle-out"], provider: { sort: "price" } },
+        },
+        headers: { "X-OpenRouter-Title": "Custom" },
+        body: { provider: { only: ["anthropic"] } },
+      }),
+      undefined,
+      {
+        loadPackage: () =>
+          Effect.succeed({
+            model: (modelID, settings) => {
+              expect(settings.headers).toEqual({
+                "HTTP-Referer": "https://opencode.ai",
+                "X-OpenRouter-Title": "Custom",
+              })
+              expect(settings.body).toEqual({
+                transforms: ["middle-out"],
+                provider: { sort: "price", only: ["anthropic"] },
+              })
+              return Model.make({ id: modelID, provider: "openrouter", route: OpenAIChat.route })
+            },
+          }),
+      },
+    ),
+  )
+
  it.effect("loads supported AISDK catalog packages as native routes", () =>
    Effect.gen(function* () {
      const google = yield* ModelResolver.fromCatalogModel(
@@ -1,291 +0,0 @@
-import { describe, expect } from "bun:test"
-import { $ } from "bun"
-import fs from "fs/promises"
-import path from "path"
-import { eq } from "drizzle-orm"
-import { Effect, Layer } from "effect"
-import { MoveSession } from "@opencode-ai/core/control-plane/move-session"
-import { Database } from "@opencode-ai/core/database/database"
-import { AppNodeBuilder } from "@opencode-ai/core/effect/app-node-builder"
-import { LayerNode } from "@opencode-ai/util/effect/layer-node"
-import { Bus } from "@opencode-ai/core/bus"
-import { Job } from "@opencode-ai/core/job"
-import { Project } from "@opencode-ai/core/project"
-import { ProjectDirectories } from "@opencode-ai/core/project/directories"
-import { AbsolutePath } from "@opencode-ai/core/schema"
-import { Session } from "@opencode-ai/core/session"
-import { SessionExecution } from "@opencode-ai/core/session/execution"
-import { SessionProjector } from "@opencode-ai/core/session/projector"
-import { SessionTable } from "@opencode-ai/core/session/sql"
-import { SessionStore } from "@opencode-ai/core/session/store"
-import { tmpdir } from "./fixture/tmpdir"
-import { testEffect } from "./lib/effect"
-
-// Records the execution serialization a move must perform before relocating.
-const executionCalls: string[] = []
-const recordingExecution = Layer.succeed(
-  SessionExecution.Service,
-  SessionExecution.Service.of({
-    active: Effect.succeed(new Set()),
-    resume: () => Effect.void,
-    wake: () => Effect.void,
-    interrupt: (sessionID) => Effect.sync(() => void executionCalls.push(`interrupt:${sessionID}`)),
-    awaitIdle: (sessionID) => Effect.sync(() => void executionCalls.push(`awaitIdle:${sessionID}`)),
-  }),
-)
-
-const it = testEffect(
-  AppNodeBuilder.build(
-    LayerNode.group([
-      MoveSession.node,
-      Database.node,
-      Bus.node,
-      ProjectDirectories.node,
-      Project.node,
-      Session.node,
-      SessionProjector.node,
-      SessionStore.node,
-    ]),
-    [[SessionExecution.node, recordingExecution]],
-  ),
-)
-
-function abs(input: string) {
-  return AbsolutePath.make(input)
-}
-
-async function initRepo(directory: string) {
-  await $`git init`.cwd(directory).quiet()
-  await $`git config core.autocrlf false`.cwd(directory).quiet()
-  await $`git config core.fsmonitor false`.cwd(directory).quiet()
-  await $`git config commit.gpgsign false`.cwd(directory).quiet()
-  await $`git config user.email test@opencode.test`.cwd(directory).quiet()
-  await $`git config user.name Test`.cwd(directory).quiet()
-  await fs.writeFile(path.join(directory, "tracked.txt"), "initial\n")
-  await $`git add tracked.txt`.cwd(directory).quiet()
-  await $`git commit -m root`.cwd(directory).quiet()
-}
-
-describe("MoveSession", () => {
-  it.live("moves session changes to another project directory", () =>
-    Effect.gen(function* () {
-      const root = yield* Effect.acquireRelease(
-        Effect.promise(() => tmpdir()),
-        (dir) => Effect.promise(() => dir[Symbol.asyncDispose]()),
-      )
-      yield* Effect.promise(() => initRepo(root.path))
-      const source = abs(yield* Effect.promise(() => fs.realpath(root.path)))
-      const destination = abs(`${root.path}-move-destination`)
-      yield* Effect.addFinalizer(() =>
-        Effect.promise(() => fs.rm(destination, { recursive: true, force: true })).pipe(Effect.ignore),
-      )
-      yield* Effect.promise(() => $`git worktree add --detach ${destination} HEAD`.cwd(root.path).quiet())
-      const moved = abs(yield* Effect.promise(() => fs.realpath(destination)))
-      yield* Effect.promise(() => fs.writeFile(path.join(source, "tracked.txt"), "changed\n"))
-      yield* Effect.promise(() => fs.writeFile(path.join(source, "untracked.txt"), "new\n"))
-
-      const projectID = (yield* Project.Service.use((service) => service.resolve(source))).id
-      const sessionID = Session.ID.make("ses_move")
-      const { db } = yield* Database.Service
-      yield* db
-        .insert(SessionTable)
-        .values({
-          id: sessionID,
-          project_id: projectID,
-          slug: "move",
-          directory: source,
-          title: "move",
-          version: "test",
-          time_created: 1,
-          time_updated: 1,
-        })
-        .run()
-        .pipe(Effect.orDie)
-
-      executionCalls.length = 0
-      yield* MoveSession.Service.use((service) =>
-        service.moveSession({ sessionID, destination: { directory: moved }, moveChanges: true }),
-      )
-
-      // The move stops active execution before any relocation side effect.
-      expect(executionCalls).toEqual([`interrupt:${sessionID}`, `awaitIdle:${sessionID}`])
-      expect(yield* Effect.promise(() => fs.readFile(path.join(moved, "tracked.txt"), "utf8"))).toBe("changed\n")
-      expect(yield* Effect.promise(() => fs.readFile(path.join(moved, "untracked.txt"), "utf8"))).toBe("new\n")
-      expect(yield* Effect.promise(() => fs.readFile(path.join(source, "tracked.txt"), "utf8"))).toBe("initial\n")
-      expect(yield* Effect.promise(() => Bun.file(path.join(source, "untracked.txt")).exists())).toBe(false)
-      expect(
-        yield* db
-          .select({ directory: SessionTable.directory, path: SessionTable.path })
-          .from(SessionTable)
-          .where(eq(SessionTable.id, sessionID))
-          .get(),
-      ).toEqual({ directory: moved, path: "" })
-    }),
-  )
-
-  it.live("moves within a checkout without transferring existing changes", () =>
-    Effect.gen(function* () {
-      const root = yield* Effect.acquireRelease(
-        Effect.promise(() => tmpdir()),
-        (dir) => Effect.promise(() => dir[Symbol.asyncDispose]()),
-      )
-      yield* Effect.promise(() => initRepo(root.path))
-      const source = abs(yield* Effect.promise(() => fs.realpath(root.path)))
-      const destination = abs(path.join(source, "packages"))
-      yield* Effect.promise(() => fs.writeFile(path.join(source, "tracked.txt"), "changed\n"))
-      yield* Effect.promise(() => fs.writeFile(path.join(source, "untracked.txt"), "new\n"))
-
-      const projectID = (yield* Project.Service.use((service) => service.resolve(source))).id
-      const sessionID = Session.ID.make("ses_move_nested")
-      const { db } = yield* Database.Service
-      yield* db
-        .insert(SessionTable)
-        .values({
-          id: sessionID,
-          project_id: projectID,
-          slug: "move-nested",
-          directory: source,
-          title: "move nested",
-          version: "test",
-          time_created: 1,
-          time_updated: 1,
-        })
-        .run()
-        .pipe(Effect.orDie)
-
-      const missing = yield* Session.Service.use((service) =>
-        service.move({ sessionID, directory: abs("packages") }).pipe(Effect.flip),
-      )
-      expect(missing._tag).toBe("Session.DestinationNotFoundError")
-      yield* Effect.promise(() => fs.mkdir(destination))
-
-      yield* MoveSession.Service.use((service) =>
-        service.moveSession({ sessionID, destination: { directory: abs("packages") }, moveChanges: true }),
-      )
-
-      expect(yield* Effect.promise(() => fs.readFile(path.join(source, "tracked.txt"), "utf8"))).toBe("changed\n")
-      expect(yield* Effect.promise(() => fs.readFile(path.join(source, "untracked.txt"), "utf8"))).toBe("new\n")
-      expect(
-        yield* db
-          .select({ directory: SessionTable.directory, path: SessionTable.path })
-          .from(SessionTable)
-          .where(eq(SessionTable.id, sessionID))
-          .get(),
-      ).toEqual({ directory: destination, path: "packages" })
-    }),
-  )
-
-  it.live("moves a session to another project", () =>
-    Effect.gen(function* () {
-      const root = yield* Effect.acquireRelease(
-        Effect.promise(() => tmpdir()),
-        (dir) => Effect.promise(() => dir[Symbol.asyncDispose]()),
-      )
-      yield* Effect.promise(() => initRepo(root.path))
-      const source = abs(yield* Effect.promise(() => fs.realpath(root.path)))
-      const destination = abs(`${root.path}-other-project`)
-      yield* Effect.acquireRelease(
-        Effect.promise(() => fs.mkdir(destination, { recursive: true })),
-        () => Effect.promise(() => fs.rm(destination, { recursive: true, force: true })),
-      )
-
-      const projectID = (yield* Project.Service.use((service) => service.resolve(source))).id
-      const destinationProjectID = (yield* Project.Service.use((service) => service.resolve(destination))).id
-      const sessionID = Session.ID.make("ses_move_project")
-      const { db } = yield* Database.Service
-      yield* db
-        .insert(SessionTable)
-        .values({
-          id: sessionID,
-          project_id: projectID,
-          slug: "move-project",
-          directory: source,
-          title: "move project",
-          version: "test",
-          time_created: 1,
-          time_updated: 1,
-        })
-        .run()
-        .pipe(Effect.orDie)
-
-      yield* Session.Service.use((service) =>
-        service.move({ sessionID, directory: destination }),
-      )
-
-      expect(
-        yield* db
-          .select({ projectID: SessionTable.project_id, directory: SessionTable.directory })
-          .from(SessionTable)
-          .where(eq(SessionTable.id, sessionID))
-          .get(),
-      ).toEqual({ projectID: destinationProjectID, directory: destination })
-    }),
-  )
-
-  it.live("moves nested session changes without cleaning unrelated files", () =>
-    Effect.gen(function* () {
-      const root = yield* Effect.acquireRelease(
-        Effect.promise(() => tmpdir()),
-        (dir) => Effect.promise(() => dir[Symbol.asyncDispose]()),
-      )
-      yield* Effect.promise(() => initRepo(root.path))
-      const source = abs(yield* Effect.promise(() => fs.realpath(root.path)))
-      const sourceDirectory = abs(path.join(source, "packages"))
-      yield* Effect.promise(() => fs.mkdir(sourceDirectory))
-      yield* Effect.promise(() => fs.writeFile(path.join(sourceDirectory, "tracked.txt"), "initial\n"))
-      yield* Effect.promise(() => fs.writeFile(path.join(sourceDirectory, "staged.txt"), "initial\n"))
-      yield* Effect.promise(() => $`git add packages/tracked.txt packages/staged.txt`.cwd(source).quiet())
-      yield* Effect.promise(() => $`git commit -m packages`.cwd(source).quiet())
-      const destination = abs(`${root.path}-move-nested-destination`)
-      yield* Effect.addFinalizer(() =>
-        Effect.promise(() => fs.rm(destination, { recursive: true, force: true })).pipe(Effect.ignore),
-      )
-      yield* Effect.promise(() => $`git worktree add --detach ${destination} HEAD`.cwd(source).quiet())
-      const moved = abs(path.join(yield* Effect.promise(() => fs.realpath(destination)), "packages"))
-      yield* Effect.promise(() => fs.writeFile(path.join(sourceDirectory, "tracked.txt"), "changed\n"))
-      yield* Effect.promise(() => fs.writeFile(path.join(sourceDirectory, "staged.txt"), "staged\n"))
-      yield* Effect.promise(() => $`git add packages/staged.txt`.cwd(source).quiet())
-      yield* Effect.promise(() => fs.writeFile(path.join(sourceDirectory, "untracked.txt"), "new\n"))
-      yield* Effect.promise(() => fs.writeFile(path.join(source, "tracked.txt"), "unrelated\n"))
-      yield* Effect.promise(() => fs.writeFile(path.join(source, "untracked.txt"), "unrelated\n"))
-
-      const projectID = (yield* Project.Service.use((service) => service.resolve(source))).id
-      const sessionID = Session.ID.make("ses_move_nested_checkout")
-      const { db } = yield* Database.Service
-      yield* db
-        .insert(SessionTable)
-        .values({
-          id: sessionID,
-          project_id: projectID,
-          slug: "move-nested-checkout",
-          directory: sourceDirectory,
-          title: "move nested checkout",
-          version: "test",
-          time_created: 1,
-          time_updated: 1,
-        })
-        .run()
-        .pipe(Effect.orDie)
-
-      yield* MoveSession.Service.use((service) =>
-        service.moveSession({ sessionID, destination: { directory: moved }, moveChanges: true }),
-      )
-
-      expect(yield* Effect.promise(() => fs.readFile(path.join(moved, "tracked.txt"), "utf8"))).toBe("changed\n")
-      expect(yield* Effect.promise(() => fs.readFile(path.join(moved, "staged.txt"), "utf8"))).toBe("staged\n")
-      expect(yield* Effect.promise(() => fs.readFile(path.join(moved, "untracked.txt"), "utf8"))).toBe("new\n")
-      expect(yield* Effect.promise(() => fs.readFile(path.join(sourceDirectory, "tracked.txt"), "utf8"))).toBe(
-        "initial\n",
-      )
-      expect(yield* Effect.promise(() => Bun.file(path.join(sourceDirectory, "untracked.txt")).exists())).toBe(false)
-      expect(yield* Effect.promise(() => fs.readFile(path.join(sourceDirectory, "staged.txt"), "utf8"))).toBe(
-        "staged\n",
-      )
-      expect(yield* Effect.promise(() => $`git status --porcelain -- packages/staged.txt`.cwd(source).text())).toBe(
-        "M  packages/staged.txt\n",
-      )
-      expect(yield* Effect.promise(() => fs.readFile(path.join(source, "tracked.txt"), "utf8"))).toBe("unrelated\n")
-      expect(yield* Effect.promise(() => fs.readFile(path.join(source, "untracked.txt"), "utf8"))).toBe("unrelated\n")
-    }),
-  )
-})
@@ -106,8 +106,10 @@ export function host(overrides: Overrides = {}): Plugin.Context {
      prompt: overrides.session?.prompt ?? (() => Effect.die("unused session.prompt")),
      generate: overrides.session?.generate ?? (() => Effect.die("unused session.generate")),
      command: overrides.session?.command ?? (() => Effect.die("unused session.command")),
+      rename: overrides.session?.rename ?? (() => Effect.die("unused session.rename")),
      synthetic: overrides.session?.synthetic ?? (() => Effect.die("unused session.synthetic")),
      interrupt: overrides.session?.interrupt ?? (() => Effect.die("unused session.interrupt")),
+      wait: overrides.session?.wait ?? (() => Effect.die("unused session.wait")),
    },
  }
 }
@@ -74,6 +74,9 @@ describe("OpenAIPlugin", () => {
          ]
        })
        catalog.model.update(item.id, Model.ID.make("gpt-5.5-pro"), () => {})
+        catalog.model.update(item.id, Model.ID.make("gpt-5.4"), (model) => {
+          model.limit = { context: 1_050_000, input: 922_000, output: 64_000 }
+        })
        catalog.model.update(item.id, Model.ID.make("gpt-5.4-pro"), (model) => {
          model.modelID = Model.ID.make("gpt-5.4")
          model.body = { reasoning: { mode: "pro" } }
@@ -137,7 +140,7 @@ describe("OpenAIPlugin", () => {
      const eligible = required(yield* catalog.model.get(Provider.ID.openai, Model.ID.make("gpt-5.5")))
      expect(eligible.package).toBe("@opencode-ai/ai/providers/openai")
      expect(eligible.cost).toEqual([])
-      expect(eligible.limit).toEqual({ context: 400_000, input: 272_000, output: 128_000 })
+      expect(eligible.limit).toEqual({ context: 272_000, input: 272_000, output: 128_000 })
      expect(eligible.enabled).toBe(true)
      expect(required(yield* catalog.model.get(Provider.ID.openai, Model.ID.make("gpt-5.5-pro"))).enabled).toBe(
        false,
@@ -145,10 +148,15 @@ describe("OpenAIPlugin", () => {
      expect(required(yield* catalog.model.get(Provider.ID.openai, Model.ID.make("gpt-5.4-pro"))).enabled).toBe(
        false,
      )
+      expect(required(yield* catalog.model.get(Provider.ID.openai, Model.ID.make("gpt-5.4"))).limit).toEqual({
+        context: 272_000,
+        input: 272_000,
+        output: 64_000,
+      })
      expect(required(yield* catalog.model.get(Provider.ID.openai, Model.ID.make("gpt-5.6"))).enabled).toBe(false)
      const gpt56 = required(yield* catalog.model.get(Provider.ID.openai, Model.ID.make("gpt-5.6-sol")))
      expect(gpt56.enabled).toBe(true)
-      expect(gpt56.limit).toEqual({ context: 500_000, input: 372_000, output: 128_000 })
+      expect(gpt56.limit).toEqual({ context: 272_000, input: 272_000, output: 128_000 })
      expect(required(yield* catalog.model.get(Provider.ID.openai, Model.ID.make("gpt-4.1"))).enabled).toBe(false)
    }),
  )
@@ -19,9 +19,11 @@ import { Session } from "@opencode-ai/core/session"
 import { Project } from "@opencode-ai/core/project"
 import { ProjectTable } from "@opencode-ai/core/project/sql"
 import { App } from "@opencode-ai/core/app"
+import { Agent } from "@opencode-ai/core/agent"
+import { Location } from "@opencode-ai/core/location"
 import { AbsolutePath } from "@opencode-ai/core/schema"
 import { Money } from "@opencode-ai/schema/money"
-import { DateTime, Effect, Fiber, Layer, Stream } from "effect"
+import { DateTime, Effect, Fiber, Layer, Schema, Stream } from "effect"
 import { asc, eq } from "drizzle-orm"
 import { testEffect } from "./lib/effect"

@@ -130,6 +132,52 @@ test("compaction prompt requires the checkpoint headings in order", () => {
  expect(prompt).toContain("Keep every section, even when empty.")
 })

+it.effect("auto compaction reserves a buffer below the prompt ceiling", () =>
+  Effect.gen(function* () {
+    const compaction = yield* SessionCompaction.Service
+    const session = Session.Info.make({
+      id: Session.ID.make("ses_input_limit"),
+      projectID: Project.ID.global,
+      cost: Money.USD.zero,
+      tokens: { input: 0, output: 0, reasoning: 0, cache: { read: 0, write: 0 } },
+      time: { created: DateTime.makeUnsafe(0), updated: DateTime.makeUnsafe(0) },
+      location: Location.Ref.make({ directory: AbsolutePath.make("/tmp") }),
+    })
+    const input = (tokens: number, limits: { context: number; input?: number; output: number }) => ({
+      session,
+      model: Model.make({
+        id: "test-model",
+        provider: "test-provider",
+        route: OpenAIChat.route.with({ limits }),
+      }),
+      cost: [],
+      messages: [
+        Schema.decodeUnknownSync(SessionMessage.Assistant)({
+          id: SessionMessage.ID.make("msg_assistant"),
+          type: "assistant",
+          agent: Agent.defaultID,
+          model: { id: "test-model", providerID: "test-provider" },
+          content: [],
+          tokens: { input: tokens, output: 0, reasoning: 0, cache: { read: 0, write: 0 } },
+          time: { created: 0, completed: 0 },
+        }),
+      ],
+    })
+
+    const inputLimited = { context: 400_000, input: 272_000, output: 128_000 }
+    expect(compaction.required(input(251_999, inputLimited))).toBe(false)
+    expect(compaction.required(input(252_000, inputLimited))).toBe(true)
+
+    const contextLimited = { context: 100_000, output: 10_000 }
+    expect(compaction.required(input(79_999, contextLimited))).toBe(false)
+    expect(compaction.required(input(80_000, contextLimited))).toBe(true)
+
+    const outputLimited = { context: 100_000, output: 30_000 }
+    expect(compaction.required(input(69_999, outputLimited))).toBe(false)
+    expect(compaction.required(input(70_000, outputLimited))).toBe(true)
+  }),
+)
+
 it.effect("manual compaction summarizes short context instead of no-op", () =>
  Effect.gen(function* () {
    requests = []
@@ -14,6 +14,9 @@ import {
  TransportReason,
  UnknownProviderReason,
  ToolFailure,
+  HttpContext,
+  HttpRequestDetails,
+  HttpResponseDetails,
 } from "@opencode-ai/ai"
 import { Permission } from "@opencode-ai/core/permission"
 import { Tool } from "@opencode-ai/schema/tool"
@@ -71,6 +74,23 @@ describe("toSessionError", () => {
    })
  })

+  test("preserves provider HTTP status", () => {
+    const http = new HttpContext({
+      request: new HttpRequestDetails({ method: "POST", url: "https://example.com", headers: {} }),
+      response: new HttpResponseDetails({ status: 413, headers: {} }),
+    })
+    expect(toSessionError(llm(new InvalidRequestReason({ message: "too large", http })))).toEqual({
+      type: "provider.invalid-request",
+      message: "too large",
+      status: 413,
+    })
+    expect(toSessionError(llm(new ProviderInternalReason({ message: "bad gateway", status: 502 })))).toEqual({
+      type: "provider.internal",
+      message: "bad gateway",
+      status: 502,
+    })
+  })
+
  test("retries only rate limits, provider-internal failures, and transport failures", () => {
    const eligible = [
      llm(new RateLimitReason({ message: "rate" })),
@@ -1,6 +1,6 @@
 import { describe, expect, test } from "bun:test"
 import { Message, ToolResultPart } from "@opencode-ai/ai"
-import { unsupportedParts } from "@opencode-ai/core/session/model-request"
+import { boundImages, unsupportedParts } from "@opencode-ai/core/session/model-request"

 const capabilities = (input: string[]) => ({ tools: true, input, output: ["text"] })

@@ -62,3 +62,51 @@ describe("SessionModelRequest.unsupportedParts", () => {
    expect(unsupportedParts([message], capabilities(["text", "image"]))[0]?.content).toEqual(message.content)
  })
 })
+
+describe("SessionModelRequest.boundImages", () => {
+  test("preserves images below the trigger", () => {
+    const messages = [Message.user({ type: "media", mediaType: "image/png", data: "aGVsbG8=" })]
+    expect(boundImages(messages)).toBe(messages)
+  })
+
+  test("replaces oldest images until the retained payload reaches the target", () => {
+    const image = "a".repeat(9 * 1024 * 1024)
+    const messages = [
+      Message.user({ type: "media", mediaType: "image/png", data: image, filename: "first.png" }),
+      Message.user({ type: "media", mediaType: "image/png", data: image, filename: "second.png" }),
+      Message.user({ type: "media", mediaType: "image/png", data: image, filename: "third.png" }),
+    ]
+    const result = boundImages(messages)
+
+    expect(result[0]?.content[0]).toMatchObject({ type: "text" })
+    expect(result[1]?.content[0]).toMatchObject({ type: "text" })
+    expect(result[2]?.content[0]).toMatchObject({ type: "media", filename: "third.png" })
+  })
+
+  test("replaces images nested in tool results", () => {
+    const image = "a".repeat(13 * 1024 * 1024)
+    const result = boundImages([
+      Message.tool(
+        ToolResultPart.make({
+          id: "call_1",
+          name: "read",
+          result: {
+            type: "content",
+            value: [
+              { type: "file", uri: `data:image/png;base64,${image}`, mime: "image/png", name: "first.png" },
+              { type: "file", uri: `data:image/png;base64,${image}`, mime: "image/png", name: "second.png" },
+            ],
+          },
+        }),
+      ),
+    ])
+
+    expect(result[0]?.content[0]).toMatchObject({
+      type: "tool-result",
+      result: {
+        type: "content",
+        value: [{ type: "text" }, { type: "file", name: "second.png" }],
+      },
+    })
+  })
+})
@@ -26,8 +26,8 @@ import { SessionPendingTable, SessionMessageTable, SessionTable } from "@opencod
 import { SessionStore } from "@opencode-ai/core/session/store"
 import { LocationServiceMap } from "@opencode-ai/core/location-service-map"
 import type { LocationServices } from "@opencode-ai/core/location-services"
+import { Image } from "@opencode-ai/core/image"
 import { testEffect } from "./lib/effect"
-import { imagePassthrough } from "./lib/image"

 const executionCalls: Session.ID[] = []
 const interruptCalls: Session.ID[] = []
@@ -58,7 +58,10 @@ const locations = Layer.effect(
    () =>
      // Attachment admission only needs the location-scoped Image service.
      // oxlint-disable-next-line typescript-eslint/no-unsafe-type-assertion
-      imagePassthrough as unknown as Layer.Layer<LocationServices>,
+      Layer.mock(Image.Service, {
+        normalize: (_resource, content) =>
+          Effect.succeed(content.content.length > 5 * 1024 * 1024 ? { ...content, content: "AA==" } : content),
+      }) as unknown as Layer.Layer<LocationServices>,
  ),
 )
 const it = testEffect(
@@ -386,6 +389,35 @@ describe("Session.prompt", () => {
    }),
  )

+  it.effect("normalizes large image content before validating persisted Base64", () =>
+    Effect.gen(function* () {
+      yield* setup
+      const session = yield* Session.Service
+      const pixel = Buffer.from(
+        "iVBORw0KGgoAAAANSUhEUgAAAAEAAAABCAQAAAC1HAwCAAAAC0lEQVR42mNk+A8AAQUBAScY42YAAAAASUVORK5CYII=",
+        "base64",
+      )
+      const bytes = Buffer.concat([pixel, Buffer.alloc(4_323_030 - pixel.length)])
+      const data = bytes.toString("base64")
+      expect(data).toHaveLength(5_764_040)
+
+      const message = yield* session.prompt({
+        sessionID,
+        text: "Inspect this image",
+        files: [{ uri: `data:image/png;base64,${data}` }],
+        resume: false,
+      })
+
+      expect(message.data.files).toEqual([
+        {
+          data: "AA==",
+          mime: "image/png",
+          source: { type: "inline" },
+        },
+      ])
+    }),
+  )
+
  it.effect("sniffs data URL content instead of trusting its declared MIME", () =>
    Effect.gen(function* () {
      yield* setup
@@ -2,7 +2,7 @@ import { beforeEach, describe, expect } from "bun:test"
 import path from "path"
 import { Effect, Exit, Layer, PlatformError, Stream } from "effect"
 import { Config } from "@opencode-ai/core/config"
-import { ConfigAttachments } from "@opencode-ai/core/config/attachments"
+import { ConfigMedia } from "@opencode-ai/core/config/media"
 import { AppNodeBuilder } from "@opencode-ai/core/effect/app-node-builder"
 import { LayerNode } from "@opencode-ai/util/effect/layer-node"
 import { FileSystem } from "@opencode-ai/core/filesystem"
@@ -432,8 +432,8 @@ describe("ReadTool", () => {
        new Config.Document({
          type: "document",
          info: new Config.Info({
-            attachments: new ConfigAttachments.Info({
-              image: new ConfigAttachments.Image({ auto_resize: false, max_width: 4 }),
+            media: new ConfigMedia.Info({
+              image: new ConfigMedia.Image({ auto_resize: false, max_width: 4 }),
            }),
          }),
        }),
@@ -475,7 +475,7 @@ describe("ReadTool", () => {
        new Config.Document({
          type: "document",
          info: new Config.Info({
-            attachments: new ConfigAttachments.Info({ image: new ConfigAttachments.Image({ max_width: 4 }) }),
+            media: new ConfigMedia.Info({ image: new ConfigMedia.Image({ max_width: 4 }) }),
          }),
        }),
      ])
@@ -514,8 +514,8 @@ describe("ReadTool", () => {
        new Config.Document({
          type: "document",
          info: new Config.Info({
-            attachments: new ConfigAttachments.Info({
-              image: new ConfigAttachments.Image({ max_base64_bytes: 1 }),
+            media: new ConfigMedia.Info({
+              image: new ConfigMedia.Image({ max_base64_bytes: 1 }),
            }),
          }),
        }),
@@ -16,6 +16,7 @@
    "@opencode-ai/core": "workspace:*",
    "@opencode-ai/session-ui": "workspace:*",
    "@opencode-ai/ui": "workspace:*",
+    "@opencode-ai/util": "workspace:*",
    "aws4fetch": "^1.0.20",
    "@pierre/diffs": "catalog:",
    "@solidjs/router": "catalog:",
@@ -4,6 +4,7 @@ import { SessionReview } from "@opencode-ai/session-ui/session-review"
 import { DataProvider } from "@opencode-ai/session-ui/context"
 import { FileComponentProvider } from "@opencode-ai/ui/context/file"
 import { WorkerPoolProvider } from "@opencode-ai/ui/context/worker-pool"
+import { withTimestampedFallback } from "@opencode-ai/util/session-title-fallback"
 import { createAsync, query, useParams } from "@solidjs/router"
 import { createMemo, createSignal, ErrorBoundary, For, Match, Show, Switch } from "solid-js"
 import { Share } from "~/core/share"
@@ -158,11 +159,7 @@ export default function () {
          const match = createMemo(() => Binary.search(data().session, data().sessionID, (s) => s.id))
          if (!match().found) throw new Error(`Session ${data().sessionID} not found`)
          const info = createMemo(() => data().session[match().index])
-          const title = createMemo(
-            () =>
-              info().title ??
-              `${info().parentID ? "Child" : "New"} session - ${new Date(info().time.created).toISOString()}`,
-          )
+          const title = createMemo(() => withTimestampedFallback(info()))
          const ogImage = createMemo(() => {
            const models = new Set<string>()
            const messages = data().message[data().sessionID] ?? []
@@ -29,7 +29,7 @@ export interface SessionHooks {

 export type SessionDomain = Pick<
  SessionApi<unknown>,
-  "create" | "get" | "prompt" | "generate" | "command" | "synthetic" | "interrupt"
+  "create" | "get" | "prompt" | "generate" | "command" | "synthetic" | "interrupt" | "rename" | "wait"
 > & {
  readonly hook: Hooks<SessionHooks>
 }
@@ -26,12 +26,27 @@ import type { JSX } from "@opentui/solid"
 import type { Store } from "solid-js/store"

 export interface Storage {
+  /**
+   * Durable JSON state: persisted to disk, survives hot reloads and TUI
+   * restarts, and stays live-synced across running TUI instances.
+   */
  store<Value extends object>(
    key: string,
    options: {
      readonly initial: Value
    },
  ): readonly [Store<Value>, (mutation: (draft: Value) => void) => Promise<void>]
+  /**
+   * Ephemeral in-memory state: survives plugin hot reloads (old and new
+   * generations share the same live store) and is gone when the TUI exits.
+   * Updates are synchronous and values need not be JSON-serializable.
+   */
+  memory<Value extends object>(
+    key: string,
+    options: {
+      readonly initial: Value
+    },
+  ): readonly [Store<Value>, (mutation: (draft: Value) => void) => void]
 }

 interface LocationCollection<Value> {
@@ -136,6 +151,9 @@ export interface SlotMap {
    readonly sessionID?: string
    readonly mode: "normal" | "shell"
  }
+  readonly "session.composer.top": {
+    readonly sessionID: string
+  }
  readonly "sidebar.content": {
    readonly sessionID: string
  }
@@ -149,6 +149,7 @@ export const makeSessionGroup = <I extends HttpApiMiddleware.AnyId, S>(sessionLo
      HttpApiEndpoint.post("session.create", "/api/session", {
        payload: Schema.Struct({
          id: Session.ID.pipe(Schema.optional),
+          title: Schema.String.pipe(Schema.optional),
          agent: Agent.ID.pipe(Schema.optional),
          model: Model.Ref.pipe(Schema.optional),
          location: Location.Ref.pipe(Schema.optional),
@@ -1,9 +1,11 @@
 export * as SessionError from "./session-error.js"

 import { Schema } from "effect"
+import { optional } from "./schema.js"

 export interface Error extends Schema.Schema.Type<typeof Error> {}
 export const Error = Schema.Struct({
  type: Schema.String,
  message: Schema.String,
+  status: Schema.Int.check(Schema.isBetween({ minimum: 100, maximum: 599 })).pipe(optional),
 }).annotate({ identifier: "Session.StructuredError" })
@@ -77,6 +77,7 @@ export const SessionHandler = HttpApiBuilder.group(Api, "server.session", (handl
            data: yield* session
              .create({
                id: ctx.payload.id,
+                title: ctx.payload.title,
                agent: ctx.payload.agent,
                model: ctx.payload.model,
                location: ctx.payload.location ?? { directory: AbsolutePath.make(process.cwd()) },
@@ -45,6 +45,7 @@ export type {
  StatefulColor,
 } from "./types.js"
 export { DEFAULT_CATEGORICAL, DEFAULT_THEME } from "./defaults.js"
+export { expandTheme } from "./expand.js"
 export { migrateV1 } from "./v1-migrate.js"
 export { resolveTheme, resolveThemeDocument, themeDecodeError } from "./resolve.js"
 export { selectTheme, selectThemeMode, supportsThemeMode, themeModes } from "./select.js"
--- a/Show More
+++ b/Show More
Author	SHA1	Message	Date
Aiden Cline	d16e7ce3c3	feat(ai): add native Bedrock Mantle support	2026-08-02 03:02:03 +00:00
Kit Langton	5d729521d3	refactor(tui): remove redundant code (#40081 )	2026-08-01 14:22:48 -04:00
Kit Langton	c391983ae0	fix(tui): discover plugins across config roots (#39988 )	2026-08-01 14:06:58 -04:00
Kit Langton	f158abd694	fix(tui): make vertical tabs responsive (#40080 )	2026-08-01 14:06:13 -04:00
Kit Langton	0e3da4b4f1	feat(tui): add vertical session tabs (#40072 )	2026-08-01 13:10:08 -04:00
Aiden Cline	003b22edda	fix: preserve provider error status (#39976 )	2026-08-01 11:06:00 -05:00
Kit Langton	d5f6c088f0	fix(tui): prevent sidebar scrollbar flash (#40056 )	2026-08-01 15:12:37 +00:00
Kit Langton	e872bd3c8f	feat(tui): prioritize favorite model search results (#40049 )	2026-08-01 14:54:46 +00:00
opencode-agent[bot]	85ea15e56d	test(tui): wait for mini prompt readiness (#39980 ) Co-authored-by: Kit Langton <kit.langton@gmail.com>	2026-08-01 01:05:18 +00:00
Kit Langton	6a543791b6	fix(tui): persist tab reorder once per drag (#39942 )	2026-07-31 20:36:00 -04:00
Kit Langton	47c6840752	fix(tui): harden session tab state hygiene (#39941 )	2026-07-31 20:35:07 -04:00
Kit Langton	14e4c3dfd7	fix(tui): ignore hidden tab close hitbox (#39940 )	2026-07-31 20:34:13 -04:00
opencode-agent[bot]	ee6577936d	refactor(tui): remove unused duration formatter (#39964 ) Co-authored-by: Kit Langton <kit.langton@gmail.com>	2026-07-31 20:32:46 -04:00
opencode-agent[bot]	8285a18f18	refactor(tui): remove unused revert diff parser (#39963 ) Co-authored-by: Kit Langton <kit.langton@gmail.com>	2026-07-31 20:32:35 -04:00
opencode-agent[bot]	cfcd8de62a	refactor(cli): remove unused warning helper (#39962 ) Co-authored-by: Kit Langton <kit.langton@gmail.com>	2026-07-31 20:32:21 -04:00
opencode-agent[bot]	c2db18c59b	refactor(tui): remove unwired file selection helper (#39961 ) Co-authored-by: Kit Langton <kit.langton@gmail.com>	2026-07-31 20:31:50 -04:00
opencode-agent[bot]	70393a1d18	Remove unused error data serializer (#39960 ) Co-authored-by: Kit Langton <kit.langton@gmail.com>	2026-07-31 20:31:22 -04:00
opencode-agent[bot]	eab59ea021	Remove unused locale helpers (#39959 ) Co-authored-by: Kit Langton <kit.langton@gmail.com>	2026-07-31 20:31:06 -04:00
opencode-agent[bot]	4569659b03	Remove unused Zed helpers (#39958 ) Co-authored-by: Kit Langton <kit.langton@gmail.com>	2026-07-31 20:30:53 -04:00
opencode-agent[bot]	b5afaed053	Remove unused config optional hook (#39957 ) Co-authored-by: Kit Langton <kit.langton@gmail.com>	2026-07-31 20:30:29 -04:00
opencode-agent[bot]	97f5fb3399	Remove unused subagent retry formatter (#39952 ) Co-authored-by: Kit Langton <kit.langton@gmail.com>	2026-07-31 20:30:20 -04:00
opencode-agent[bot]	18843682ad	refactor(tui): remove ignored attention kv option (#39956 ) Co-authored-by: Kit Langton <kit.langton@gmail.com>	2026-07-31 20:30:09 -04:00
opencode-agent[bot]	9eb9c1db7d	refactor(tui): remove placeholder LSP panel (#39955 ) Co-authored-by: Kit Langton <kit.langton@gmail.com>	2026-07-31 20:29:54 -04:00
opencode-agent[bot]	8846336e11	refactor(tui): remove unused fade file path (#39954 ) Co-authored-by: Kit Langton <kit.langton@gmail.com>	2026-07-31 20:29:34 -04:00
opencode-agent[bot]	f6b5f6b6b3	refactor(tui): remove prompt re-export barrels (#39953 ) Co-authored-by: Kit Langton <kit.langton@gmail.com>	2026-07-31 20:29:27 -04:00
opencode-agent[bot]	02df1e9737	refactor(tui): reuse canonical logo (#39951 ) Co-authored-by: Kit Langton <kit.langton@gmail.com>	2026-07-31 20:28:55 -04:00
opencode-agent[bot]	5b84e39f6d	refactor(tui): remove stub child cycling commands (#39949 ) Co-authored-by: Kit Langton <kit.langton@gmail.com>	2026-07-31 20:28:23 -04:00
opencode-agent[bot]	6d3f84b4ff	refactor(tui): remove dialog show helpers (#39948 ) Co-authored-by: Kit Langton <kit.langton@gmail.com>	2026-07-31 20:28:02 -04:00
opencode-agent[bot]	255e9bbf06	refactor(tui): remove onboarding hint residue (#39947 ) Co-authored-by: Kit Langton <kit.langton@gmail.com>	2026-07-31 20:27:52 -04:00
opencode-agent[bot]	25b6a179e2	refactor(cli): remove no-op migrate command (#39946 ) Co-authored-by: Kit Langton <kit.langton@gmail.com>	2026-07-31 20:27:29 -04:00
opencode-agent[bot]	80874c241f	refactor(tui): remove dead assistant renderer (#39945 ) Co-authored-by: Kit Langton <kit.langton@gmail.com>	2026-07-31 20:27:15 -04:00
opencode-agent[bot]	f99469bdf8	refactor(core): remove unused layer exports (#39975 ) Co-authored-by: Kit Langton <kit.langton@gmail.com>	2026-07-31 20:26:38 -04:00
opencode-agent[bot]	fe0eb4ea8f	refactor(core): remove orphaned move service (#39974 ) Co-authored-by: Kit Langton <kit.langton@gmail.com>	2026-07-31 20:26:21 -04:00
opencode-agent[bot]	a484f87680	refactor(core): remove unused dependencies (#39973 ) Co-authored-by: Kit Langton <kit.langton@gmail.com>	2026-07-31 20:26:06 -04:00
opencode-agent[bot]	3b6ab392ac	refactor(core): remove unused console state model (#39972 ) Co-authored-by: Kit Langton <kit.langton@gmail.com>	2026-07-31 20:25:45 -04:00
opencode-agent[bot]	6c567dc745	refactor(core): remove unreferenced layer map example (#39971 ) Co-authored-by: Kit Langton <kit.langton@gmail.com>	2026-07-31 20:25:27 -04:00
opencode-agent[bot]	92567700f6	refactor(tui): remove unused stop voice operation (#39969 ) Co-authored-by: Kit Langton <kit.langton@gmail.com>	2026-07-31 20:25:17 -04:00
James Long	90c84639e3	feat(theme): export expandTheme (#39967 )	2026-07-31 18:32:55 -04:00
opencode-agent[bot]	56c6add5c3	refactor(core): remove unused helpers (#39943 ) Co-authored-by: Aiden Cline <rekram1-node@users.noreply.github.com>	2026-07-31 16:17:14 -05:00
Kit Langton	2cfe8883ea	feat(plugin): expose session rename and wait to plugins (#39932 )	2026-07-31 21:08:33 +00:00
Kit Langton	102086c50f	feat(tui): hot-reload local TUI plugins (#39776 )	2026-07-31 16:35:05 -04:00
Kit Langton	06e26d89ad	feat(protocol): accept title on session create (#39935 )	2026-07-31 19:39:54 +00:00
Aiden Cline	84dd56ed34	fix(core): bound outbound image history (#39929 )	2026-07-31 13:57:04 -05:00
opencode-agent[bot]	f92e490bb6	fix(tui): preserve diff hunk boundaries (#39693 ) Co-authored-by: James Long <17031+jlongster@users.noreply.github.com>	2026-07-31 14:47:36 -04:00
Kit Langton	0481dba88e	fix(tui): improve narrow layouts (#39918 )	2026-07-31 14:44:13 -04:00
Kit Langton	dd0e41c633	fix(tui): render plugin slots and routes with component semantics (#39917 )	2026-07-31 14:30:57 -04:00
opencode-agent[bot]	b69e51d835	refactor(ai): simplify provider options (#39924 ) Co-authored-by: Aiden Cline <63023139+rekram1-node@users.noreply.github.com>	2026-07-31 13:27:32 -05:00
opencode-agent[bot]	e60e8d9387	refactor(core): rename attachments config to media (#39927 ) Co-authored-by: Aiden Cline <rekram1-node@users.noreply.github.com>	2026-07-31 13:26:53 -05:00
Kit Langton	3bfce3fd2d	feat(tui): expand pasted text (#39920 )	2026-07-31 18:17:35 +00:00
Aiden Cline	b498a5c6c4	feat(ai): expand OpenRouter native support (#39907 )	2026-07-31 12:57:18 -05:00
Kit Langton	db9e942398	fix(core): resize large image attachments (#39919 )	2026-07-31 13:24:05 -04:00
Kit Langton	0e26116f68	fix(tui): stabilize open picker	2026-07-31 16:38:48 +00:00
Aiden Cline	3468aa0140	feat(ai): configure chat max tokens field (#39909 )	2026-07-31 11:14:08 -05:00
Kit Langton	7757f712a2	test(app): restore stable offset observer ordering (#39901 )	2026-07-31 12:13:16 -04:00
Aiden Cline	7129ed6e62	fix(core): respect model input limits (#39797 )	2026-07-31 10:41:36 -05:00
Kit Langton	6c37842520	fix(cli): deduplicate Node asset destinations (#39900 )	2026-07-31 11:40:38 -04:00
Kit Langton	dc3c996892	fix(tui): stabilize generated session titles (#39894 )	2026-07-31 11:13:25 -04:00
Kit Langton	7fd12c560c	fix(tui): fit project paths in open menu (#39887 )	2026-07-31 11:07:12 -04:00
Kit Langton	7aaf4e7750	refactor(session): centralize fallback title policy (#39890 )	2026-07-31 10:56:38 -04:00
Kit Langton	1311d909a7	fix(tui): preserve session location during handoff (#39886 )	2026-07-31 10:54:36 -04:00
Kit Langton	d07d9ae0da	fix(tui): remove shells from their location (#39885 )	2026-07-31 10:46:43 -04:00
Kit Langton	1d18459cdc	fix(tui): align session picker scope (#39784 )	2026-07-31 10:32:16 -04:00
Kit Langton	db45026c6c	fix(tui): default tabs to global scope (#39783 )	2026-07-31 10:15:53 -04:00
Kit Langton	77df98db51	Revert "feat(tui): delete current session (#39750 )" This reverts commit `7814568ba0`.	2026-07-31 10:01:35 -04:00