Compare commits

...

7 Commits

Author SHA1 Message Date
Dustin Deus 54b8ea622c Merge branch 'dev' into kimi-mfjs-schema 2026-07-18 20:42:27 +02:00
starptech bc3686b8df fix(opencode): improve MFJS schema sanitizing 2026-07-18 20:38:32 +02:00
starptech d2a11baa99 fix(provider): harden mfjs schema projection 2026-07-18 20:04:22 +02:00
starptech 50a29ff19f Update mfjs.ts 2026-07-18 19:15:02 +02:00
starptech 4d6f3f002c chore(opencode): add tool schema compatibility matrix 2026-07-18 19:14:52 +02:00
starptech fa041090f7 fix(provider 2026-07-18 17:08:52 +02:00
starptech ba5dd2f962 fix(provider): normalize kimi tool schemas for mfjs 2026-07-18 13:22:34 +02:00
5 changed files with 1315 additions and 152 deletions
@@ -0,0 +1,285 @@
import { ModelV2 } from "@opencode-ai/core/model"
import { ProviderV2 } from "@opencode-ai/core/provider"
import type { JSONSchema7 } from "@ai-sdk/provider"
import { Effect } from "effect"
import { jsonSchema, streamText, tool, type LanguageModel } from "ai"
import { AppRuntime } from "@/effect/app-runtime"
import { InstanceRef } from "@/effect/instance-ref"
import { Provider } from "@/provider/provider"
import { InstanceStore } from "@/project/instance-store"
import { MFJS } from "@/provider/mfjs"
// Usage:
// bun run script/tool-schema-compatibility-matrix.ts --models=model-a,model-b
// [--provider=opencode-go] [--projection=none|mfjs]
// [--cases=tuple items,...] [--concurrency=9] [--timeout=30000]
type JsonRecord = Record<string, unknown>
type Case = {
name: string
schema: JsonRecord
}
type Result = {
case: string
model: string
status: "accepted" | "rejected" | "rate-limited" | "error"
code?: number
error?: string
}
const projection = option("projection") ?? "none"
if (projection !== "none" && projection !== "mfjs") throw new Error(`Unsupported projection: ${projection}`)
const concurrency = Number(option("concurrency") ?? 9)
const timeout = Number(option("timeout") ?? 30_000)
const providerID = option("provider") ?? "opencode-go"
const models = option("models")?.split(",").filter(Boolean) ?? []
if (models.length === 0) throw new Error("--models must contain at least one model ID")
const selectedCases = new Set(option("cases")?.split(",").filter(Boolean) ?? [])
const matrix: Case[] = [
property("enum/type mismatch", { type: "object", enum: ["move", "copy"] }),
property("untyped enum", { enum: ["move", "copy"] }),
property("mixed untyped enum", { enum: ["move", 1, null, true] }),
property("const", { const: "move" }),
property("tuple items", { type: "array", items: [{ type: "string" }, { type: "number" }] }),
property("prefix items", { type: "array", prefixItems: [{ type: "string" }, { type: "number" }] }),
property("typed anyOf", {
type: "string",
enum: ["move"],
anyOf: [{ type: "string" }, { type: "null" }],
}),
property("anyOf count limit", {
anyOf: Array.from({ length: 501 }, (_, index) => ({ const: `value_${index}` })),
}),
property("oneOf", { oneOf: [{ type: "string" }, { type: "integer" }] }),
property("allOf", {
allOf: [
{ type: "object", properties: { left: { type: "string" } } },
{ type: "object", properties: { right: { type: "number" } } },
],
}),
property("not", { type: "string", not: { enum: ["blocked"] } }),
property("if/then/else", {
type: "object",
properties: { mode: { type: "string" }, count: { type: "integer" } },
if: { properties: { mode: { enum: ["many"] } } },
then: { required: ["count"] },
else: { properties: { count: { maximum: 1 } } },
}),
property("contains", { type: "array", contains: { type: "string" } }),
objectCase("patternProperties", {
type: "object",
patternProperties: { "^extra_": { type: "number" } },
additionalProperties: false,
}),
objectCase("dependentSchemas", {
type: "object",
properties: { key: { type: "string" }, value: { type: "string" } },
dependentSchemas: { key: { required: ["value"] } },
}),
objectCase("propertyNames", {
type: "object",
propertyNames: { pattern: "^[a-z]+$" },
}),
property("uniqueItems", { type: "array", items: { type: "string" }, uniqueItems: true }),
property("boolean true schema", true),
property("boolean false schema", false),
objectCase("empty property name", {
type: "object",
properties: { "": { type: "string" } },
required: [""],
}),
objectCase("dangling required", {
type: "object",
properties: {},
required: ["missing"],
}),
objectCase("external ref", {
type: "object",
properties: { value: { $ref: "https://example.com/schema.json" } },
}),
objectCase("chained ref", {
type: "object",
properties: { value: { $ref: "#/$defs/A" } },
$defs: { A: { $ref: "#/$defs/B" }, B: { type: "string" } },
}),
objectCase("recursive ref", {
type: "object",
properties: { node: { $ref: "#/$defs/Node" } },
$defs: {
Node: {
type: "object",
properties: { value: { type: "string" }, next: { anyOf: [{ $ref: "#/$defs/Node" }, { type: "null" }] } },
required: ["value"],
},
},
}),
objectCase("reference depth limit", {
type: "object",
properties: { value: { $ref: "#/$defs/Value" } },
$defs: { Value: nested(30) },
}),
objectCase("schema size limit", {
type: "object",
description: "x".repeat(120_001),
properties: {},
}),
objectCase("schema depth limit", nested(35)),
objectCase("property count limit", {
type: "object",
properties: Object.fromEntries(
Array.from({ length: 3001 }, (_, index) => [`property_${index}`, { type: "string" }]),
),
}),
property("enum count limit", {
type: "string",
enum: Array.from({ length: 1001 }, (_, index) => `value_${index}`),
}),
]
const cases = selectedCases.size === 0 ? matrix : matrix.filter((item) => selectedCases.has(item.name))
const { store, ctx } = await AppRuntime.runPromise(
InstanceStore.Service.use((store) =>
store.load({ directory: process.cwd() }).pipe(Effect.map((ctx) => ({ store, ctx }))),
),
)
try {
const languages = await AppRuntime.runPromise(
Effect.gen(function* () {
const provider = yield* Provider.Service
return yield* Effect.forEach(
models,
Effect.fnUntraced(function* (model) {
const info = yield* provider.getModel(ProviderV2.ID.make(providerID), ModelV2.ID.make(model))
return [model, yield* provider.getLanguage(info)] as const
}),
{ concurrency: "unbounded" },
)
}).pipe(Effect.provideService(InstanceRef, ctx)),
)
const jobs = languages.flatMap(([model, language]) => cases.map((item) => () => run(language, model, item)))
const results = await parallel(jobs, concurrency)
print(results)
if (projection !== "none" && results.some((result) => result.status !== "accepted")) process.exitCode = 1
} finally {
await AppRuntime.runPromise(store.dispose(ctx))
}
process.exit(process.exitCode ?? 0)
async function run(language: LanguageModel, model: string, item: Case): Promise<Result> {
const schema = projection === "mfjs" ? MFJS.sanitize(item.schema) : item.schema
let providerError: unknown
try {
const response = streamText({
model: language,
prompt: "Reply with exactly OK without calling tools.",
maxOutputTokens: 16,
abortSignal: AbortSignal.timeout(timeout),
onError(event) {
providerError = event.error
},
tools: {
probe: tool({
description: `Tool schema probe: ${item.name}`,
inputSchema: jsonSchema(schema as JSONSchema7),
execute: async () => "ok",
}),
},
})
await response.text
if (providerError) throw providerError
console.error(`accepted: ${model} / ${item.name}`)
return { case: item.name, model, status: "accepted" }
} catch (error) {
const failure = providerError ?? error
const code = statusCode(failure)
const message = failure instanceof Error ? failure.message : String(failure)
const status = code === 429 || /rate limit/i.test(message) ? "rate-limited" : code === 400 ? "rejected" : "error"
console.error(`${status}: ${model} / ${item.name}`)
return {
case: item.name,
model,
status,
code,
error: message,
}
}
}
async function parallel<T>(jobs: Array<() => Promise<T>>, limit: number) {
const output = new Array<T>(jobs.length)
let next = 0
await Promise.all(
Array.from({ length: Math.min(limit, jobs.length) }, async () => {
while (true) {
const index = next++
const job = jobs[index]
if (!job) return
output[index] = await job()
}
}),
)
return output
}
function print(results: Result[]) {
const byCase = new Map<string, Map<string, Result>>()
results.forEach((result) => {
const row = byCase.get(result.case) ?? new Map<string, Result>()
row.set(result.model, result)
byCase.set(result.case, row)
})
console.log(`Projection: ${projection}`)
console.log(`Provider: ${providerID}`)
console.log(`| Case | ${models.join(" | ")} |`)
console.log(`| --- | ${models.map(() => "---").join(" | ")} |`)
cases.forEach((item) => {
const row = byCase.get(item.name)
const values = models.map((model) => {
const result = row?.get(model)
if (!result) return "missing"
return result.status === "accepted" ? "accepted" : `${result.status}${result.code ? ` (${result.code})` : ""}`
})
console.log(`| ${item.name} | ${values.join(" | ")} |`)
})
const rejected = results.filter((result) => result.status !== "accepted")
if (rejected.length > 0) {
console.log("\nRejected details:")
rejected.forEach((result) => console.log(`- ${result.model} / ${result.case}: ${result.error}`))
}
}
function property(name: string, schema: unknown): Case {
return objectCase(name, { type: "object", properties: { value: schema }, required: ["value"] })
}
function objectCase(name: string, schema: JsonRecord): Case {
return { name, schema }
}
function nested(depth: number): JsonRecord {
return Array.from({ length: depth }).reduce<JsonRecord>(
(schema) => ({ type: "object", properties: { next: schema }, required: ["next"] }),
{ type: "string" },
)
}
function option(name: string) {
const prefix = `--${name}=`
return process.argv.find((arg) => arg.startsWith(prefix))?.slice(prefix.length)
}
function statusCode(error: unknown, seen = new Set<object>()): number | undefined {
if (!isRecord(error) || seen.has(error)) return
seen.add(error)
if (typeof error.statusCode === "number") return error.statusCode
for (const value of Object.values(error)) {
const nested = statusCode(value, seen)
if (nested !== undefined) return nested
}
}
function isRecord(value: unknown): value is JsonRecord {
return typeof value === "object" && value !== null && !Array.isArray(value)
}
+651
View File
@@ -0,0 +1,651 @@
export * as MFJS from "./mfjs"
import type { JSONSchema7 } from "@ai-sdk/provider"
/**
* Kimi tool-schema compatibility projection.
*
* Principles:
* - Preserve schema features accepted by Kimi without rewriting them.
* - Apply model-agnostic adaptations only for reproduced provider rejections.
* - Keep explicit types authoritative; lossy fallbacks may widen but never narrow.
*
* Adapted families include enum/type conflicts, untyped enums, tuple `items`,
* typed `anyOf`, boolean schemas, dangling `required`, references, and observed
* enum/union/property/size/depth limits. Accepted keywords such as `const`,
* `oneOf`, `allOf`, conditionals, `prefixItems`, and other constraints pass
* through recursively.
*
* `script/tool-schema-compatibility-matrix.ts` compares raw and MFJS-projected
* schemas across configured providers and models. Keep this projection
* evidence-driven as provider behavior evolves.
*
* MFJS specification and reference implementation:
* https://github.com/MoonshotAI/walle
*/
type JsonRecord = Record<string, unknown>
type Context = {
root: JsonRecord
definitions: JsonRecord
legacy: Record<string, string>
properties: number
}
type Projection = {
schema: JsonRecord
// Unsafe projections cannot stay under non-monotonic applicators without risking narrowing.
unsafe: boolean
}
const TYPES = new Set(["string", "number", "boolean", "integer", "object", "array", "null"])
const SCHEMA_MAPS = new Set(["patternProperties", "dependentSchemas"])
const SCHEMA_NODES = new Set([
"additionalProperties",
"not",
"if",
"then",
"else",
"contains",
"propertyNames",
"unevaluatedProperties",
])
const SCHEMA_LISTS = new Set(["allOf", "prefixItems"])
const MAX_ANY_OF = 500
const MAX_DEPTH = 30
const MAX_ENUM = 1000
const MAX_PROPERTIES = 3000
const MAX_RECURSION = 1000
const MAX_SCHEMA_SIZE = 120_000
/**
* Projects tool schemas only where Kimi rejects otherwise valid requests.
* Accepted keywords are preserved, explicit types remain authoritative, and
* lossy fallbacks only widen what the model may emit.
*/
export function sanitize(value: unknown): JSONSchema7 {
const root = isRecord(value) ? value : {}
const sourceDefinitions = definitions(root)
const context = { root, definitions: sourceDefinitions.schemas, legacy: sourceDefinitions.legacy, properties: 0 }
const projected = project(root, context, 0, 0).schema
const defs = Object.fromEntries(
Object.entries(context.definitions)
.filter(([name, schema]) => name.length > 0 && !name.includes("/") && isRecord(schema))
.map(([name, schema]) => [name, containsSlashKey(schema) ? {} : project(schema, context, 0, 0).schema]),
)
if (Object.keys(defs).length > 0) projected.$defs = defs
const resolved = dropDanglingRefs(projected, projected)
const bounded =
(containsRef(resolved) && !terminates(resolved, resolved, new Set())) ||
schemaDepth(resolved, resolved, new Set()) > MAX_DEPTH
? { type: "object", properties: {} }
: resolved
return fitSize(bounded) as JSONSchema7
}
function project(value: unknown, context: Context, depth: number, recursion: number): Projection {
if (depth >= MAX_DEPTH || recursion >= MAX_RECURSION) return { schema: {}, unsafe: true }
if (!isRecord(value) || Object.keys(value).length === 0) {
return { schema: {}, unsafe: !isRecord(value) }
}
const ref = canonicalRef(value.$ref, context)
if (ref) {
const schema = value.nullable === true ? { anyOf: [{ $ref: ref }, { type: "null" }] } : { $ref: ref }
return {
schema,
// References are conservative here because their projected targets may widen later.
unsafe: true,
}
}
const declaredTypes = schemaTypes(value.type)
const inferredTypes =
declaredTypes.length > 0
? declaredTypes
: hasUnprojectableEnumValue("const" in value ? [value.const] : value.enum, [])
? []
: groupEnum(enumValues("const" in value ? [value.const] : value.enum)).map((group) => group.type)
if (Array.isArray(value.anyOf) && inferredTypes.length > 0) {
return projectTypedAnyOf(
{ ...value, type: inferredTypes.length === 1 ? inferredTypes[0] : inferredTypes },
context,
depth,
recursion,
)
}
const result: JsonRecord = {}
let unsafe = "$ref" in value || "$defs" in value || "definitions" in value
let truncatedProperties = false
let widenedContains = false
const child = (item: unknown, nextDepth = depth, nextContext = context) =>
project(item, nextContext, nextDepth, recursion + 1)
const keep = (projection: Projection) => {
unsafe ||= projection.unsafe
return projection.schema
}
const condition = (() => {
if (!isRecord(value.if)) return
const nested = { ...context }
const projection = child(value.if, depth, nested)
if (!projection.unsafe) context.properties = nested.properties
return projection
})()
for (const [key, item] of Object.entries(value)) {
if (key === "$defs" || key === "definitions") continue
if (key === "$ref") continue
if (key === "items" && Array.isArray(item)) {
unsafe = true
continue
}
if (key === "items" && (item === true || item === false)) {
result.items = {}
unsafe = true
continue
}
if (key === "anyOf" && Array.isArray(item)) {
if (item.length > 0 && item.length <= MAX_ANY_OF) {
result.anyOf = item.map((branch) => keep(child(branch)))
} else {
unsafe = true
}
continue
}
if (key === "oneOf" && Array.isArray(item)) {
const nested = { ...context }
const branches = item.map((branch) => child(branch, depth, nested))
const risky = branches.some((branch) => branch.unsafe)
const useBranches =
!risky || (!Array.isArray(value.anyOf) && branches.length > 0 && branches.length <= MAX_ANY_OF)
if (!risky) result.oneOf = branches.map((branch) => branch.schema)
if (risky && useBranches) result.anyOf = branches.map((branch) => branch.schema)
if (useBranches) context.properties = nested.properties
unsafe ||= risky
continue
}
if (SCHEMA_LISTS.has(key) && Array.isArray(item)) {
result[key] = item.map((schema) => keep(child(schema)))
continue
}
if (key === "not" && isRecord(item)) {
const nested = { ...context }
const schema = child(item, depth, nested)
if (!schema.unsafe) {
result.not = schema.schema
context.properties = nested.properties
} else {
unsafe = true
}
continue
}
if (key === "if" || key === "then" || key === "else") {
if (condition?.unsafe) {
unsafe = true
continue
}
if (key === "if" && condition) {
result.if = condition.schema
continue
}
if (isRecord(item)) {
result[key] = keep(child(item))
continue
}
result[key] = item
continue
}
if (key === "properties" && isRecord(item)) {
const remaining = Math.max(0, MAX_PROPERTIES - context.properties)
const entries = Object.entries(item).slice(0, remaining)
context.properties += entries.length
truncatedProperties = entries.length !== Object.keys(item).length
result.properties = Object.fromEntries(entries.map(([name, schema]) => [name, keep(child(schema, depth + 1))]))
unsafe ||= truncatedProperties
continue
}
if (SCHEMA_MAPS.has(key) && isRecord(item)) {
const schemas = Object.entries(item).map(([name, schema]) => [name, child(schema)] as const)
result[key] = Object.fromEntries(
schemas.map(([name, schema]) => [name, typeof schema.schema.$ref === "string" ? {} : schema.schema]),
)
unsafe ||= schemas.some(([, schema]) => schema.unsafe)
continue
}
if (key === "contains" && isRecord(item)) {
const schema = child(item)
result.contains = schema.schema
widenedContains = schema.unsafe
keep(schema)
continue
}
if ((key === "items" || SCHEMA_NODES.has(key)) && isRecord(item)) {
result[key] = keep(child(item))
continue
}
result[key] = item
}
unsafe ||= projectRequired(result, context)
if (truncatedProperties) delete result.additionalProperties
if (widenedContains && "maxContains" in result) {
delete result.maxContains
unsafe = true
}
const projected = projectEnum(result)
unsafe ||= projected.unsafe
if ("unevaluatedProperties" in projected.schema && unsafe) {
delete projected.schema.unevaluatedProperties
unsafe = true
}
return { schema: projected.schema, unsafe }
}
function projectTypedAnyOf(source: JsonRecord, context: Context, depth: number, recursion: number): Projection {
const base = omit(source, ["anyOf", "type", "enum", "const", "$defs", "definitions"])
const parentTypes = schemaTypes(source.type)
const parentValues = "const" in source ? [source.const] : source.enum
const parentEnum = hasUnprojectableEnumValue(parentValues, parentTypes) ? [] : enumValues(parentValues)
const variants = Array.isArray(source.anyOf) ? source.anyOf : []
if (variants.length > MAX_ANY_OF) {
const projected = project(omit(source, ["anyOf", "unevaluatedProperties"]), context, depth, recursion + 1)
return { ...projected, unsafe: true }
}
const branches = variants.flatMap((branch) => {
if (branch === false) return []
const item = branch === true ? {} : isRecord(branch) ? branch : undefined
if (!item) return []
const types = intersectTypes(parentTypes, schemaTypes(item.type))
if (types.length === 0) return []
const branchValues = "const" in item ? [item.const] : item.enum
const branchEnum = hasUnprojectableEnumValue(branchValues, types) ? [] : enumValues(branchValues)
const values = intersectEnums(parentEnum, branchEnum).filter((value) =>
types.some((type) => matchesType(value, type)),
)
if ((parentEnum.length > 0 || branchEnum.length > 0) && values.length === 0) return []
const merged = omit({ ...base, ...item }, ["type", "enum", "const"])
const projected = project(
{
...merged,
type: types.length === 1 ? types[0] : types,
...(values.length > 0 ? { enum: values } : {}),
},
context,
depth,
recursion + 1,
)
return [projected.schema]
})
return { schema: collapse(branches), unsafe: true }
}
function projectEnum(source: JsonRecord): Projection {
const result = { ...source }
const types = schemaTypes(result.type)
if (hasUnprojectableEnumValue(result.enum, types)) {
delete result.enum
return { schema: result, unsafe: true }
}
const values = enumValues(result.enum)
if (values.length === 0) {
const unsafe = "enum" in result
delete result.enum
return { schema: result, unsafe }
}
if (types.length > 0) {
const compatible = values.filter((value) => types.some((type) => matchesType(value, type)))
if (compatible.length === 0) {
delete result.enum
return { schema: result, unsafe: true }
}
const nullable =
types.length === 2 && types.includes("null") && !types.includes("object") && !types.includes("array")
if (types.length === 1 || nullable) {
result.enum = compatible
return { schema: result, unsafe: !same(result.enum, source.enum) }
}
const base = omit(result, ["enum", "type"])
return {
schema: collapse(groupEnum(compatible).map((group) => ({ ...base, type: group.type, enum: group.values }))),
unsafe: true,
}
}
const groups = groupEnum(values)
if (groups.length === 1) {
result.type = groups[0]?.type
result.enum = groups[0]?.values
return { schema: result, unsafe: true }
}
const base = omit(result, ["enum", "type"])
return {
schema: { anyOf: groups.map((group) => ({ ...base, type: group.type, enum: group.values })) },
unsafe: true,
}
}
function projectRequired(schema: JsonRecord, context: Context) {
if (!Array.isArray(schema.required)) return false
const types = schemaTypes(schema.type)
if (types.length > 0 && !types.includes("object")) {
delete schema.required
return true
}
const sourceProperties = isRecord(schema.properties) ? schema.properties : undefined
const properties = { ...sourceProperties }
const sourceRequired = schema.required
const required = [...new Set(schema.required.filter((item): item is string => typeof item === "string"))].filter(
(name) => {
if (Object.hasOwn(properties, name)) return true
if (context.properties >= MAX_PROPERTIES) return false
context.properties++
properties[name] = {}
return true
},
)
schema.properties = properties
schema.required = required
return (
!sourceProperties ||
!same(Object.keys(sourceProperties), Object.keys(properties)) ||
!same(sourceRequired, required)
)
}
function fitSize(schema: JsonRecord) {
if (schemaSize(schema) <= MAX_SCHEMA_SIZE) return schema
const compact = stripAnnotations(schema)
if (schemaSize(compact) <= MAX_SCHEMA_SIZE) return compact
return {}
}
function stripAnnotations(schema: JsonRecord): JsonRecord {
return Object.fromEntries(
Object.entries(schema).flatMap(([key, value]) => {
if (["description", "title", "default", "examples", "$comment"].includes(key)) return []
if ((key === "properties" || key === "$defs" || SCHEMA_MAPS.has(key)) && isRecord(value)) {
return [
[
key,
Object.fromEntries(
Object.entries(value).map(([name, item]) => [name, isRecord(item) ? stripAnnotations(item) : item]),
),
],
]
}
if ((key === "anyOf" || key === "oneOf" || SCHEMA_LISTS.has(key)) && Array.isArray(value)) {
return [[key, value.map((item) => (isRecord(item) ? stripAnnotations(item) : item))]]
}
if ((key === "items" || SCHEMA_NODES.has(key)) && isRecord(value)) {
return [[key, stripAnnotations(value)]]
}
return [[key, value]]
}),
)
}
function schemaSize(schema: JsonRecord) {
const json = JSON.stringify(schema).replace(/[<>&\u2028\u2029]/g, (char) => {
return `\\u${char.charCodeAt(0).toString(16).padStart(4, "0")}`
})
return new TextEncoder().encode(json).byteLength
}
function schemaDepth(schema: JsonRecord, root: JsonRecord, refs: Set<string>): number {
const properties = isRecord(schema.properties)
? Math.max(
0,
...Object.values(schema.properties).map((item) => (isRecord(item) ? 1 + schemaDepth(item, root, refs) : 0)),
)
: 0
const nodes = [schema.items, schema.additionalProperties].flatMap((item) =>
isRecord(item) ? [schemaDepth(item, root, refs)] : [],
)
const lists = [schema.anyOf, schema.oneOf, schema.allOf, schema.prefixItems].flatMap((items) =>
Array.isArray(items) ? items.flatMap((item) => (isRecord(item) ? [schemaDepth(item, root, refs)] : [])) : [],
)
const definitions = isRecord(schema.$defs)
? Object.values(schema.$defs).flatMap((item) => (isRecord(item) ? [schemaDepth(item, root, refs)] : []))
: []
const ref = (() => {
if (typeof schema.$ref !== "string" || refs.has(schema.$ref)) return 0
const target = resolveOutputRef(schema.$ref, root)
if (!target) return 0
const next = new Set(refs)
next.add(schema.$ref)
return schemaDepth(target, root, next)
})()
return [...nodes, ...lists, ...definitions, ref].reduce((max, value) => Math.max(max, value), properties)
}
function dropDanglingRefs(schema: JsonRecord, root: JsonRecord): JsonRecord {
if (typeof schema.$ref === "string" && !resolveOutputRef(schema.$ref, root)) return {}
return Object.fromEntries(
Object.entries(schema).map(([key, value]) => {
if ((key === "properties" || key === "$defs" || SCHEMA_MAPS.has(key)) && isRecord(value)) {
return [
key,
Object.fromEntries(
Object.entries(value).map(([name, item]) => [name, isRecord(item) ? dropDanglingRefs(item, root) : item]),
),
]
}
if ((key === "anyOf" || key === "oneOf" || SCHEMA_LISTS.has(key)) && Array.isArray(value)) {
return [key, value.map((item) => (isRecord(item) ? dropDanglingRefs(item, root) : item))]
}
if ((key === "items" || SCHEMA_NODES.has(key)) && isRecord(value)) {
return [key, dropDanglingRefs(value, root)]
}
return [key, value]
}),
)
}
function resolveOutputRef(ref: string, root: JsonRecord): JsonRecord | undefined {
if (ref === "#") return root
const defs = isRecord(root.$defs) ? root.$defs : undefined
if (!ref.startsWith("#/$defs/") || !defs) return
return ref
.slice("#/$defs/".length)
.split("/")
.map((part) => part.replaceAll("~1", "/").replaceAll("~0", "~"))
.reduce<JsonRecord | undefined>((current, part, index) => {
const value = index === 0 ? defs[part] : current?.[part]
return isRecord(value) ? value : undefined
}, undefined)
}
function terminates(schema: JsonRecord, root: JsonRecord, refs: Set<string>): boolean {
const types = schemaTypes(schema.type)
if (types.some((type) => type !== "object" && type !== "array")) return true
if (types.includes("array")) {
if (!isRecord(schema.items) || Object.keys(schema.items).length === 0) return true
if (terminates(schema.items, root, refs)) return true
}
if (types.includes("object")) {
if (!Array.isArray(schema.required) || schema.required.length === 0) return true
const properties = isRecord(schema.properties) ? schema.properties : undefined
if (!properties || Object.keys(properties).length === 0) return true
if (
schema.required.every((name) => {
const property = typeof name === "string" ? properties[name] : undefined
return !isRecord(property) || terminates(property, root, refs)
})
) {
return true
}
}
if (
Array.isArray(schema.anyOf) &&
(schema.anyOf.length === 0 || schema.anyOf.some((item) => isRecord(item) && terminates(item, root, new Set(refs))))
) {
return true
}
if (typeof schema.$ref === "string") {
if (refs.has(schema.$ref)) return false
const target = resolveOutputRef(schema.$ref, root)
if (!target) return true
const next = new Set(refs)
next.add(schema.$ref)
return terminates(target, root, next)
}
return Object.keys(schema).length === 0
}
function canonicalRef(value: unknown, context: Context) {
if (typeof value !== "string") return
if (value.includes("~0") || value.includes("~1")) return
const ref = (() => {
if (!value.startsWith("#/definitions/")) return value
const parts = value.slice("#/definitions/".length).split("/")
const name = parts[0]?.replaceAll("~1", "/").replaceAll("~0", "~")
if (!name || !context.legacy[name]) return
return `#/$defs/${context.legacy[name]?.replaceAll("~", "~0").replaceAll("/", "~1")}${
parts.length > 1 ? `/${parts.slice(1).join("/")}` : ""
}`
})()
if (!ref) return
if (ref === "#") return ref
if (!ref.startsWith("#/$defs/") || ref === "#/$defs/") return
const name = ref.slice("#/$defs/".length).split("/", 1)[0]?.replaceAll("~1", "/").replaceAll("~0", "~")
if (!name || name.includes("/")) return
return isRecord(resolveRef(ref, context)) ? ref : undefined
}
function resolveRef(ref: string, context: Context): unknown {
if (ref === "#") return context.root
return ref
.slice("#/$defs/".length)
.split("/")
.map((part) => part.replaceAll("~1", "/").replaceAll("~0", "~"))
.reduce<unknown>((current, part, index) => {
if (index === 0) return context.definitions[part]
return isRecord(current) ? current[part] : undefined
}, undefined)
}
function definitions(root: JsonRecord) {
const modern = isRecord(root.$defs) ? root.$defs : {}
const schemas = { ...modern }
const legacy = Object.fromEntries(
Object.entries(isRecord(root.definitions) ? root.definitions : {}).map(([name, schema]) => {
const target =
!Object.hasOwn(schemas, name) || same(schemas[name], schema) ? name : uniqueDefinition(name, schemas)
schemas[target] = schema
return [name, target]
}),
)
return { schemas, legacy }
}
function uniqueDefinition(name: string, schemas: JsonRecord, index = 1): string {
const candidate = `${name}__definitions${index === 1 ? "" : `_${index}`}`
if (!Object.hasOwn(schemas, candidate)) return candidate
return uniqueDefinition(name, schemas, index + 1)
}
function containsSlashKey(value: unknown): boolean {
if (Array.isArray(value)) return value.some(containsSlashKey)
if (!isRecord(value)) return false
return Object.entries(value).some(([key, item]) => key.includes("/") || containsSlashKey(item))
}
function schemaTypes(value: unknown) {
if (typeof value === "string" && TYPES.has(value)) return [value]
if (!Array.isArray(value)) return []
return [...new Set(value.filter((item): item is string => typeof item === "string" && TYPES.has(item)))]
}
function intersectTypes(parent: string[], child: string[]) {
if (child.length === 0) return parent
return [
...new Set([
...parent.filter((type) => child.includes(type)),
...((parent.includes("number") && child.includes("integer")) ||
(parent.includes("integer") && child.includes("number"))
? ["integer"]
: []),
]),
]
}
function enumValues(value: unknown) {
if (!Array.isArray(value)) return []
const values = unique(value.filter((item) => valueType(item) !== undefined))
return values.length > MAX_ENUM ? [] : values
}
function hasUnprojectableEnumValue(value: unknown, types: string[]) {
if (!Array.isArray(value)) return false
return value.some((item) => {
if (valueType(item)) return false
const type = Array.isArray(item) ? "array" : isRecord(item) ? "object" : undefined
return type !== undefined && (types.length === 0 || types.includes(type))
})
}
function intersectEnums(parent: unknown[], child: unknown[]) {
if (parent.length === 0) return child
if (child.length === 0) return parent
return parent.filter((value) => child.some((item) => Object.is(value, item)))
}
function groupEnum(values: unknown[]) {
const hasDecimal = values.some((item) => typeof item === "number" && !Number.isInteger(item))
return values.reduce<{ type: string; values: unknown[] }[]>((groups, item) => {
const actual = valueType(item)
if (!actual) return groups
const type = hasDecimal && actual === "integer" ? "number" : actual
const group = groups.find((entry) => entry.type === type)
if (group) group.values.push(item)
else groups.push({ type, values: [item] })
return groups
}, [])
}
function valueType(value: unknown) {
if (value === null) return "null"
if (typeof value === "string" || typeof value === "boolean") return typeof value
if (typeof value === "number" && Number.isFinite(value)) return Number.isInteger(value) ? "integer" : "number"
}
function matchesType(value: unknown, type: string) {
const actual = valueType(value)
if (type === "number") return actual === "number" || actual === "integer"
return actual === type
}
function collapse(branches: JsonRecord[]) {
const projected = unique(branches)
if (projected.length === 0) return {}
if (projected.length === 1) return projected[0] ?? {}
return { anyOf: projected }
}
function unique<T>(values: T[]) {
return [...new Map(values.map((value) => [JSON.stringify(value), value])).values()]
}
function same(left: unknown, right: unknown) {
return JSON.stringify(left) === JSON.stringify(right)
}
function containsRef(value: unknown): boolean {
if (!isRecord(value)) return false
if (typeof value.$ref === "string") return true
const maps = [value.properties, value.$defs, value.definitions, ...[...SCHEMA_MAPS].map((key) => value[key])]
if (maps.some((map) => isRecord(map) && Object.values(map).some(containsRef))) return true
const nodes = [value.items, ...[...SCHEMA_NODES].map((key) => value[key])]
if (nodes.some(containsRef)) return true
return [value.anyOf, value.oneOf, ...[...SCHEMA_LISTS].map((key) => value[key])].some(
(items) => Array.isArray(items) && items.some(containsRef),
)
}
function omit(source: JsonRecord, keys: string[]) {
const omitted = new Set(keys)
return Object.fromEntries(Object.entries(source).filter(([key]) => !omitted.has(key)))
}
function isRecord(value: unknown): value is JsonRecord {
return typeof value === "object" && value !== null && !Array.isArray(value)
}
+7 -16
View File
@@ -2,6 +2,7 @@ import type { ModelMessage, ToolResultPart } from "ai"
import { mergeDeep, unique } from "remeda"
import type { JSONSchema7 } from "@ai-sdk/provider"
import type * as Provider from "./provider"
import { MFJS } from "./mfjs"
import type * as ModelsDev from "@opencode-ai/core/models-dev"
import { iife } from "@/util/iife"
@@ -1462,22 +1463,12 @@ export function schema(model: Provider.Model, schema: JSONSchema7): JSONSchema7
// Codex also applies lossy compaction above 4 KB; defer that until OpenCode needs the same schema budget.
}
if (model.providerID === "moonshotai" || model.api.id.toLowerCase().includes("kimi")) {
const sanitizeMoonshot = (obj: unknown): unknown => {
if (obj === null || typeof obj !== "object") return obj
if (Array.isArray(obj)) return obj.map(sanitizeMoonshot)
// Moonshot expands $ref before validation and rejects sibling keywords like description on the same node.
if ("$ref" in obj && typeof obj.$ref === "string") return { $ref: obj.$ref }
const result = Object.fromEntries(Object.entries(obj).map(([key, value]) => [key, sanitizeMoonshot(value)]))
// MFJS does not support tuple-style `items` arrays; it requires one schema object for all array items.
if (Array.isArray(result.items)) result.items = result.items[0] ?? {}
return result
}
const sanitized = sanitizeMoonshot(schema)
if (typeof sanitized === "object" && sanitized !== null && !Array.isArray(sanitized)) {
schema = sanitized
}
if (
model.providerID === "moonshotai" ||
model.family?.toLowerCase().startsWith("kimi") ||
model.api.id.toLowerCase().includes("kimi")
) {
schema = MFJS.sanitize(schema)
}
// Convert integer enums to string enums for Google/Gemini
@@ -0,0 +1,354 @@
import { describe, expect, test } from "bun:test"
import { MFJS } from "@/provider/mfjs"
function expectProjection(input: unknown, expected: unknown) {
expect(JSON.stringify(MFJS.sanitize(input))).toBe(JSON.stringify(expected))
}
function asObject(value: unknown) {
if (typeof value === "object" && value !== null && !Array.isArray(value)) return value as Record<string, unknown>
throw new Error("expected object")
}
describe("MFJS.sanitize", () => {
test("removes reference siblings and normalizes draft-07 definitions", () => {
expect(
MFJS.sanitize({
type: "object",
properties: {
value: { $ref: "#/definitions/Value", description: "drop me" },
},
definitions: {
Value: { type: "object", description: "keep me" },
},
}),
).toEqual({
type: "object",
properties: { value: { $ref: "#/$defs/Value" } },
$defs: { Value: { type: "object", description: "keep me" } },
})
})
test("preserves nullable reference semantics", () => {
expect(
MFJS.sanitize({
type: "object",
properties: { value: { $ref: "#/$defs/Value", nullable: true } },
$defs: { Value: { type: "object" } },
}),
).toEqual({
type: "object",
properties: { value: { anyOf: [{ $ref: "#/$defs/Value" }, { type: "null" }] } },
$defs: { Value: { type: "object" } },
})
})
test("keeps explicit types and removes incompatible enum values", () => {
expect(
MFJS.sanitize({
type: "object",
properties: { operation: { type: "object", enum: ["move", "copy"] } },
required: ["operation"],
}),
).toEqual({
type: "object",
properties: { operation: { type: "object" } },
required: ["operation"],
})
})
test("removes only enum values excluded by an explicit type", () => {
expect(
MFJS.sanitize({
type: "object",
properties: { operation: { type: "string", enum: ["move", null] } },
}),
).toEqual({
type: "object",
properties: { operation: { type: "string", enum: ["move"] } },
})
})
test("infers a type for homogeneous untyped enums", () => {
expect(
MFJS.sanitize({
type: "object",
properties: { operation: { enum: ["move", "copy"] } },
}),
).toEqual({
type: "object",
properties: { operation: { type: "string", enum: ["move", "copy"] } },
})
})
test("splits mixed untyped enums into homogeneous branches", () => {
expect(
MFJS.sanitize({
type: "object",
properties: { value: { enum: ["move", 1] } },
}),
).toEqual({
type: "object",
properties: {
value: {
anyOf: [
{ type: "string", enum: ["move"] },
{ type: "integer", enum: [1] },
],
},
},
})
})
test("widens enums when structured values remain possible", () => {
expect(
MFJS.sanitize({
type: "object",
properties: {
untyped: { enum: ["text", { kind: "legacy" }] },
excluded: { type: "string", enum: ["text", { kind: "legacy" }] },
union: {
type: ["string", "object"],
enum: ["text", { kind: "legacy" }],
anyOf: [{ type: "string" }, { type: "object" }],
},
},
}),
).toEqual({
type: "object",
properties: {
untyped: {},
excluded: { type: "string", enum: ["text"] },
union: { anyOf: [{ type: "string" }, { type: "object" }] },
},
})
})
test("drops tuple items instead of narrowing positional schemas", () => {
expect(
MFJS.sanitize({
type: "object",
properties: {
values: {
type: "array",
items: [{ type: "string" }, { type: "number" }],
minItems: 2,
},
},
}),
).toEqual({
type: "object",
properties: { values: { type: "array", minItems: 2 } },
})
})
test("preserves prefixItems accepted by Kimi", () => {
expectProjection(
{
type: "object",
properties: {
values: { type: "array", prefixItems: [{ type: "string" }, { type: "number" }] },
},
},
{
type: "object",
properties: {
values: { type: "array", prefixItems: [{ type: "string" }, { type: "number" }] },
},
},
)
})
test("moves parent types into compatible anyOf branches", () => {
expect(
MFJS.sanitize({
type: "object",
properties: {
value: {
type: "string",
enum: ["move"],
anyOf: [{ type: "string" }, { type: "null" }],
},
},
}),
).toEqual({
type: "object",
properties: { value: { type: "string", enum: ["move"] } },
})
})
test("preserves type arrays and nullable accepted by Kimi", () => {
expectProjection(
{
type: "object",
properties: {
typed: { type: ["string", "null"], enum: ["move", null] },
nullable: { type: "string", nullable: true },
},
},
{
type: "object",
properties: {
typed: { type: ["string", "null"], enum: ["move", null] },
nullable: { type: "string", nullable: true },
},
},
)
})
test("preserves oneOf, allOf, and other keywords accepted by Kimi", () => {
const value = {
oneOf: [
{ type: "string", format: "uri" },
{ type: "integer", multipleOf: 2 },
],
examples: ["https://example.com"],
}
const all = {
allOf: [
{ type: "object", properties: { left: { type: "string" } } },
{ type: "object", properties: { right: { type: "number" } } },
],
}
expectProjection(
{ type: "object", properties: { value, all }, unevaluatedProperties: false },
{ type: "object", properties: { value, all }, unevaluatedProperties: false },
)
})
test("preserves const accepted by Kimi", () => {
expect(
MFJS.sanitize({
type: "object",
properties: { operation: { const: "move" } },
}),
).toEqual({
type: "object",
properties: { operation: { const: "move" } },
})
})
test("drops unresolved references", () => {
expect(
MFJS.sanitize({
type: "object",
properties: { value: { $ref: "#/definitions/Missing" } },
}),
).toEqual({
type: "object",
properties: { value: {} },
})
})
test("drops recursive schemas with no finite instance", () => {
expect(
MFJS.sanitize({
type: "object",
properties: { node: { $ref: "#/$defs/Node" } },
required: ["node"],
$defs: {
Node: {
type: "object",
properties: {
value: { type: "string" },
next: { $ref: "#/$defs/Node" },
},
required: ["value", "next"],
},
},
}),
).toEqual({ type: "object", properties: {} })
})
test("adds unconstrained schemas for dangling required properties", () => {
expect(
MFJS.sanitize({
type: "object",
properties: {},
required: ["missing"],
}),
).toEqual({
type: "object",
properties: { missing: {} },
required: ["missing"],
})
})
test("preserves unconstrained schemas", () => {
expect(
MFJS.sanitize({
type: "object",
properties: { empty: {}, truthy: true, falsy: false },
}),
).toEqual({
type: "object",
properties: { empty: {}, truthy: {}, falsy: {} },
})
})
test("widens schemas that exceed Kimi limits", () => {
const properties = Object.fromEntries(
Array.from({ length: 3001 }, (_, index) => [`property_${index}`, { type: "string" }]),
)
const limited = asObject(MFJS.sanitize({ type: "object", properties }))
expect(Object.keys(asObject(limited.properties))).toHaveLength(3000)
expect(
MFJS.sanitize({
type: "object",
properties: {
value: { type: "string", enum: Array.from({ length: 1001 }, (_, index) => `value_${index}`) },
},
}),
).toEqual({
type: "object",
properties: { value: { type: "string" } },
})
expect(
MFJS.sanitize({
type: "object",
properties: {
value: { anyOf: Array.from({ length: 501 }, (_, index) => ({ const: `value_${index}` })) },
},
}),
).toEqual({
type: "object",
properties: { value: {} },
})
const deep = Array.from({ length: 35 }).reduce<Record<string, unknown>>(
(schema) => ({ type: "object", properties: { next: schema }, required: ["next"] }),
{ type: "string" },
)
expect(JSON.stringify(MFJS.sanitize(deep)).match(/properties/g)?.length).toBe(30)
expect(MFJS.sanitize({ type: "object", description: "<".repeat(20_000), properties: {} })).toEqual({
type: "object",
properties: {},
})
const definition = Array.from({ length: 30 }).reduce<Record<string, unknown>>(
(schema) => ({ type: "object", properties: { next: schema } }),
{ type: "string" },
)
expect(
MFJS.sanitize({
type: "object",
properties: { value: { $ref: "#/$defs/Value" } },
$defs: { Value: definition },
}),
).toEqual({ type: "object", properties: {} })
})
test("is idempotent", () => {
const once = MFJS.sanitize({
type: "object",
properties: {
operation: { type: "object", enum: ["move"] },
values: { type: "array", items: [{ type: "string" }, { type: "number" }] },
},
})
expect(MFJS.sanitize(once)).toEqual(once)
})
})
+18 -136
View File
@@ -1528,148 +1528,30 @@ describe("ProviderTransform.schema - openai supported schema subset", () => {
})
})
describe("ProviderTransform.schema - moonshot $ref siblings", () => {
const moonshotModel = {
providerID: "moonshotai",
api: {
id: "kimi-k2",
},
} as any
describe("ProviderTransform.schema - MFJS selection", () => {
const models = [
["Moonshot providers", { providerID: "moonshotai", api: { id: "kimi-k2" } }],
["Kimi API IDs", { providerID: "openrouter", api: { id: "moonshotai/kimi-k2" } }],
["Kimi model families", { providerID: "custom", family: "kimi-k2", api: { id: "alias" } }],
] as const
test("removes sibling descriptions from referenced tool parameter schemas", () => {
const schema = {
type: "object",
properties: {
deviceType: {
description: "Optional. The type of device that captured the screenshot, e.g. mobile or desktop.",
enum: ["DEVICE_TYPE_UNSPECIFIED", "MOBILE", "DESKTOP", "TABLET", "AGNOSTIC"],
type: "string",
},
modelId: {
description: "Optional. The model to use for generation.",
enum: ["MODEL_ID_UNSPECIFIED", "GEMINI_3_PRO", "GEMINI_3_FLASH", "GEMINI_3_1_PRO"],
type: "string",
},
projectId: {
description: "Required. The project ID of screens to generate variants for.",
type: "string",
},
prompt: {
description: "Required. The input text used to generate the variants.",
type: "string",
},
selectedScreenIds: {
description: "Required. The screen ids of screen to generate variants for.",
items: {
type: "string",
},
type: "array",
},
variantOptions: {
$ref: "#/$defs/VariantOptions",
description:
"Required. The variant options for generation, including the number of variants, creative range, and aspects to focus on.",
},
},
required: ["projectId", "selectedScreenIds", "prompt", "variantOptions"],
$defs: {
VariantOptions: {
description:
"Configuration options for design variant generation. This message captures all parameters used to generate variants, allowing the configuration to be stored, replayed, or analyzed.",
properties: {
aspects: {
description: "Optional. Specific aspects to focus on. If empty, all aspects may be varied.",
items: {
enum: ["VARIANT_ASPECT_UNSPECIFIED", "LAYOUT", "COLOR_SCHEME", "IMAGES", "TEXT_FONT", "TEXT_CONTENT"],
type: "string",
},
type: "array",
},
creativeRange: {
description: "Optional. Creative range for variations. Default: EXPLORE",
enum: ["CREATIVE_RANGE_UNSPECIFIED", "REFINE", "EXPLORE", "REIMAGINE"],
type: "string",
},
variantCount: {
description: "Optional. Number of variants to generate (1-5). Default: 3",
format: "int32",
type: "integer",
},
},
for (const [name, model] of models) {
test(`sanitizes ${name}`, () => {
expect(
ProviderTransform.schema(model as Parameters<typeof ProviderTransform.schema>[0], {
type: "object",
},
},
description: "Request message for GenerateVariants.",
additionalProperties: false,
} as any
const result = ProviderTransform.schema(moonshotModel, schema) as any
expect(result.properties.variantOptions).toEqual({
$ref: "#/$defs/VariantOptions",
})
expect(result.$defs.VariantOptions.description).toBe(schema.$defs.VariantOptions.description)
})
test("also runs for kimi models outside the moonshot provider", () => {
const result = ProviderTransform.schema(
{
providerID: "openrouter",
name: "Kimi K2",
api: {
id: "moonshotai/kimi-k2",
},
} as any,
{
properties: {
operation: { type: "object", enum: ["move", "copy"] },
},
}),
).toEqual({
type: "object",
properties: {
value: {
$ref: "#/$defs/Value",
description: "Moonshot rejects this sibling after ref expansion.",
},
operation: { type: "object" },
},
$defs: {
Value: {
description: "Referenced schema description stays here.",
type: "object",
},
},
} as any,
) as any
expect(result.properties.value).toEqual({
$ref: "#/$defs/Value",
})
})
})
test("converts tuple-style array items to a single item schema", () => {
const result = ProviderTransform.schema(moonshotModel, {
type: "object",
properties: {
codeSpec: {
type: "object",
properties: {
accessibility: {
type: "object",
properties: {
renderedSize: {
description: "Rendered size [width, height] in px",
type: "array",
items: [{ type: "number" }, { type: "number" }],
minItems: 2,
maxItems: 2,
},
},
},
},
},
},
} as any) as any
expect(result.properties.codeSpec.properties.accessibility.properties.renderedSize.items).toEqual({
type: "number",
})
})
}
})
describe("ProviderTransform.message - DeepSeek reasoning content", () => {