@tanstack/ai 0.42.0 → 0.43.1

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@tanstack/ai",
-  "version": "0.42.0",
+  "version": "0.43.1",
   "description": "Type-safe TypeScript AI SDK for streaming chat, tool calling, agents, structured outputs, and multimodal generation.",
   "author": "Tanner Linsley",
   "license": "MIT",
@@ -29,6 +29,10 @@
       "types": "./dist/esm/client.d.ts",
       "import": "./dist/esm/client.js"
     },
+    "./locks": {
+      "types": "./dist/esm/locks.d.ts",
+      "import": "./dist/esm/locks.js"
+    },
     "./adapters": {
       "types": "./dist/esm/activities/index.d.ts",
       "import": "./dist/esm/activities/index.js"
@@ -73,11 +77,11 @@
     "tanstack-intent"
   ],
   "dependencies": {
-    "@ag-ui/core": "^0.0.52",
+    "@ag-ui/core": "0.1.1-canary.beta.0",
     "@standard-schema/spec": "^1.1.0",
     "partial-json": "^0.1.7",
-    "@tanstack/ai-event-client": "0.6.8",
-    "@tanstack/ai-utils": "0.3.1"
+    "@tanstack/ai-event-client": "0.7.0",
+    "@tanstack/ai-utils": "0.4.0"
   },
   "peerDependencies": {
     "@opentelemetry/api": ">=1.9.0"
@@ -96,11 +100,11 @@
   "scripts": {
     "build": "vite build",
     "clean": "premove ./build ./dist",
-    "lint:fix": "eslint ./src --fix",
+    "lint:fix": "oxlint src --type-aware --fix",
     "test:build": "publint --strict",
     "test:coverage": "vitest run --coverage",
     "test:coverage:watch": "vitest --coverage --watch",
-    "test:eslint": "eslint ./src",
+    "test:oxlint": "oxlint src --type-aware",
     "test:lib": "vitest",
     "test:lib:dev": "pnpm test:lib --watch",
     "test:types": "tsc"

package/skills/ai-core/SKILL.md

@@ -3,12 +3,13 @@ name: ai-core
 description: >
   Entry point for TanStack AI skills. Routes to chat-experience, tool-calling,
   media-generation, structured-outputs, adapter-configuration, ag-ui-protocol,
-  middleware, custom-backend-integration, and debug-logging. Use chat() not
-  streamText(), openaiText() not createOpenAI(), toServerSentEventsResponse()
-  not manual SSE, middleware hooks not onEnd callbacks.
+  middleware, locks, custom-backend-integration, and debug-logging, plus the skills
+  shipped by companion packages (@tanstack/ai-persistence, @tanstack/ai-code-mode).
+  Use chat() not streamText(), openaiText() not createOpenAI(),
+  toServerSentEventsResponse() not manual SSE, middleware hooks not onEnd callbacks.
 type: core
 library: tanstack-ai
-library_version: '0.10.0'
+library_version: '0.42.0'
 ---
 
 # TanStack AI — Core Concepts
@@ -21,18 +22,67 @@ Always import from the framework package on the client — never from
 
 ## Sub-Skills
 
-| Need to...                                        | Read                                        |
-| ------------------------------------------------- | ------------------------------------------- |
-| Build a chat UI with streaming                    | ai-core/chat-experience/SKILL.md            |
-| Add tool calling (server, client, or both)        | ai-core/tool-calling/SKILL.md               |
-| Generate images, video, speech, or transcriptions | ai-core/media-generation/SKILL.md           |
-| Get typed JSON responses from the LLM             | ai-core/structured-outputs/SKILL.md         |
-| Choose and configure a provider adapter           | ai-core/adapter-configuration/SKILL.md      |
-| Implement AG-UI streaming protocol server-side    | ai-core/ag-ui-protocol/SKILL.md             |
-| Add analytics, logging, or lifecycle hooks        | ai-core/middleware/SKILL.md                 |
-| Connect to a non-TanStack-AI backend              | ai-core/custom-backend-integration/SKILL.md |
-| Turn on/off debug logging, pipe into pino/winston | ai-core/debug-logging/SKILL.md              |
-| Set up Code Mode (LLM code execution)             | See `@tanstack/ai-code-mode` package skills |
+| Need to...                                        | Read                                          |
+| ------------------------------------------------- | --------------------------------------------- |
+| Build a chat UI with streaming                    | ai-core/chat-experience/SKILL.md              |
+| Survive a browser reload (no extra package)       | ai-core/client-persistence/SKILL.md           |
+| Add tool calling (server, client, or both)        | ai-core/tool-calling/SKILL.md                 |
+| Generate images, video, speech, or transcriptions | ai-core/media-generation/SKILL.md             |
+| Get typed JSON responses from the LLM             | ai-core/structured-outputs/SKILL.md           |
+| Choose and configure a provider adapter           | ai-core/adapter-configuration/SKILL.md        |
+| Implement AG-UI streaming protocol server-side    | ai-core/ag-ui-protocol/SKILL.md               |
+| Add analytics, logging, or lifecycle hooks        | ai-core/middleware/SKILL.md                   |
+| Coordinate multi-instance work with locks         | ai-core/locks/SKILL.md                        |
+| Connect to a non-TanStack-AI backend              | ai-core/custom-backend-integration/SKILL.md   |
+| Turn on/off debug logging, pipe into pino/winston | ai-core/debug-logging/SKILL.md                |
+| Persist chats server-side (history, runs)         | See `@tanstack/ai-persistence` package skills |
+| Set up Code Mode (LLM code execution)             | See `@tanstack/ai-code-mode` package skills   |
+
+## Companion packages
+
+Some capabilities live in their own package and ship their own skills. Install
+the package, then read its skills — do not guess the API from this file.
+
+### `@tanstack/ai-persistence` — durable chat state
+
+Makes a conversation survive a reload, a server restart, a second device, or a
+paused tool approval. It ships the **store contracts** (`MessageStore`,
+`RunStore`, `InterruptStore`, `MetadataStore`), the `withPersistence` /
+`withGenerationPersistence` middleware, `reconstructChat` for server-side
+hydrate, an in-memory reference backend, and a conformance testkit. Multi-instance
+locks are **not** in this package — `LockStore` / `withLocks` ship in
+`@tanstack/ai/locks`; see ai-core/locks. The `runs` store contract is typed
+against run lifecycle types (`RunStatus`, `RunRecord`, `RunStore`,
+`defineRunStore`, `InMemoryRunStore`), which ship in `@tanstack/ai` itself;
+see ai-core/middleware.
+
+It does **not** ship a backend for your database — you implement the stores
+against Postgres, SQLite, D1, Mongo, or whatever you run, and the package's
+skills walk you through it (including Drizzle, Prisma, and Cloudflare recipes).
+
+```bash
+pnpm add @tanstack/ai-persistence
+npx @tanstack/intent@latest install
+```
+
+The skills ship **inside** the package, so they only exist on disk once it is
+installed — the second command re-scans `node_modules` and wires them into the
+agent config. Until then the paths below resolve to nothing.
+
+Entry point: `node_modules/@tanstack/ai-persistence/skills/ai-persistence/SKILL.md`
+
+| Need to...                                      | Read                                    |
+| ----------------------------------------------- | --------------------------------------- |
+| Wire server-side chat history, runs, interrupts | ai-persistence/server/SKILL.md          |
+| Implement the store interfaces for your DB      | ai-persistence/stores/SKILL.md          |
+| Write the adapter for the DB your app runs      | ai-persistence/build-*-adapter/SKILL.md |
+
+Browser-side persistence is **not** in this package — it ships with the
+framework packages, so read **ai-core/client-persistence** instead.
+
+### `@tanstack/ai-code-mode` — LLM code execution
+
+See the `ai-code-mode` skill in that package.
 
 ## Quick Decision Tree
 
@@ -43,6 +93,7 @@ Always import from the framework package on the client — never from
 - Choosing/configuring a provider? → ai-core/adapter-configuration
 - Building a server-only AG-UI backend? → ai-core/ag-ui-protocol
 - Adding analytics or post-stream events? → ai-core/middleware
+- Surviving reloads / multi-device / durable approvals? → `@tanstack/ai-persistence` skills
 - Connecting to a custom backend? → ai-core/custom-backend-integration
 - Turning on debug logging to trace chunks/tools/middleware? → ai-core/debug-logging
 - Debugging mistakes? → Check Common Mistakes in the relevant sub-skill
@@ -54,8 +105,8 @@ Always import from the framework package on the client — never from
 3. **Use `toServerSentEventsResponse()`** to convert streams to HTTP responses. Never implement SSE manually.
 4. **Use middleware for lifecycle events.** No `onEnd`/`onFinish` callbacks on `chat()` — use `middleware: [{ onFinish: ... }]`.
 5. **Ask the user which adapter and model** they want. Suggest the latest model. Also ask if they want Code Mode.
-6. **Tools must be passed to both server and client.** Server gets the tool in `chat({ tools })`, client gets the definition in `useChat({ clientTools })`.
+6. **Tools must be passed to both server and client.** Server gets the tool in `chat({ tools })`; the client passes the `.client()` implementation through the `clientTools()` helper into the **`tools`** option — `useChat({ tools: clientTools(myTool.client(...)) })`. There is no `clientTools` option. See ai-core/tool-calling.
 
 ## Version
 
-Targets TanStack AI v0.10.0.
+Targets TanStack AI v0.42.0.

package/skills/ai-core/adapter-configuration/SKILL.md

@@ -3,16 +3,19 @@ name: ai-core/adapter-configuration
 description: >
   Provider adapter selection and configuration: openaiText, anthropicText,
   geminiText, ollamaText, grokText, groqText, openRouterText, bedrockText,
-  openaiCompatible. Per-model type safety with modelOptions, reasoning/thinking configuration,
+  byteplusText, openaiCompatible. Per-model type safety with modelOptions,
+  reasoning/thinking configuration,
   runtime adapter switching, extendAdapter() for custom models, createModel().
   Generic OpenAI-compatible providers (DeepSeek, Together, Fireworks, etc.) via
   openaiCompatible({ baseURL, apiKey, models }) from @tanstack/ai-openai/compatible.
   API key env vars: OPENAI_API_KEY, ANTHROPIC_API_KEY, GOOGLE_API_KEY/GEMINI_API_KEY,
   XAI_API_KEY, GROQ_API_KEY, OPENROUTER_API_KEY, OLLAMA_HOST,
-  BEDROCK_API_KEY (or AWS_BEARER_TOKEN_BEDROCK).
+  BEDROCK_API_KEY (or AWS_BEARER_TOKEN_BEDROCK). BytePlus needs TWO keys:
+  ARK_API_KEY (ModelArk — chat/video/image) and BYTEPLUS_VOICE_API_KEY
+  (Seed Speech — TTS/transcription); neither is a fallback for the other.
 type: sub-skill
 library: tanstack-ai
-library_version: '0.10.0'
+library_version: '0.42.0'
 sources:
   - 'TanStack/ai:docs/adapters/openai.md'
   - 'TanStack/ai:docs/adapters/anthropic.md'
@@ -80,8 +83,15 @@ The text adapter is the primary one for chat/completions:
 | OpenRouter        | `@tanstack/ai-openrouter`        | `openRouterText`                            | `OPENROUTER_API_KEY`                              |
 | Ollama            | `@tanstack/ai-ollama`            | `ollamaText`                                | `OLLAMA_HOST` (default: `http://localhost:11434`) |
 | Bedrock           | `@tanstack/ai-bedrock`           | `bedrockText`                               | `BEDROCK_API_KEY` or `AWS_BEARER_TOKEN_BEDROCK`   |
+| BytePlus          | `@tanstack/ai-byteplus`          | `byteplusText`                              | `ARK_API_KEY` (falls back to `BYTEPLUS_API_KEY`)  |
 | OpenAI-compatible | `@tanstack/ai-openai/compatible` | `openaiCompatible` / `openaiCompatibleText` | provider-specific (passed via `apiKey`)           |
 
+> **BytePlus uses two keys.** `byteplusText` / `byteplusVideo` /
+> `byteplusImage` read `ARK_API_KEY` (ModelArk, `Authorization: Bearer`), but
+> `byteplusSpeech` / `byteplusTranscription` are a separate product and read
+> **`BYTEPLUS_VOICE_API_KEY`** (Seed Speech, `X-Api-Key`). Ark keys are also
+> region-isolated — the default base URL is the ap-southeast endpoint.
+
 ```typescript
 // Each factory takes model as first arg, optional config as second
 import { openaiText } from '@tanstack/ai-openai'
@@ -92,6 +102,7 @@ import { groqText } from '@tanstack/ai-groq'
 import { openRouterText } from '@tanstack/ai-openrouter'
 import { ollamaText } from '@tanstack/ai-ollama'
 import { bedrockText } from '@tanstack/ai-bedrock'
+import { byteplusText } from '@tanstack/ai-byteplus'
 
 // Model string is passed to the factory, NOT to chat()
 const adapter = openaiText('gpt-5.2')
@@ -102,6 +113,7 @@ const adapter5 = groqText('llama-3.3-70b-versatile')
 const adapter6 = openRouterText('anthropic/claude-sonnet-4')
 const adapter7 = ollamaText('llama3.3')
 const adapter8 = bedrockText('us.anthropic.claude-3-7-sonnet-20250219-v1:0')
+const adapter9 = byteplusText('seed-2-0-lite-260428')
 
 // Optional: pass explicit API key
 const adapterWithKey = openaiText('gpt-5.2', {
@@ -283,15 +295,16 @@ chat({
 
 Per-provider sampling keys (all live inside `modelOptions`):
 
-| Provider          | Temperature   | Nucleus | Max output tokens                   |
-| ----------------- | ------------- | ------- | ----------------------------------- |
-| OpenAI            | `temperature` | `top_p` | `max_output_tokens`                 |
-| Anthropic         | `temperature` | `top_p` | `max_tokens`                        |
-| Gemini            | `temperature` | `topP`  | `maxOutputTokens`                   |
-| Grok (xAI)        | `temperature` | `top_p` | `max_tokens`                        |
-| Groq              | `temperature` | `top_p` | `max_completion_tokens`             |
-| OpenRouter (chat) | `temperature` | `topP`  | `maxCompletionTokens`               |
-| Ollama            | `temperature` | `top_p` | `num_predict` (nested in `options`) |
+| Provider          | Temperature   | Nucleus | Max output tokens                         |
+| ----------------- | ------------- | ------- | ----------------------------------------- |
+| OpenAI            | `temperature` | `top_p` | `max_output_tokens`                       |
+| Anthropic         | `temperature` | `top_p` | `max_tokens`                              |
+| Gemini            | `temperature` | `topP`  | `maxOutputTokens`                         |
+| Grok (xAI)        | `temperature` | `top_p` | `max_tokens`                              |
+| Groq              | `temperature` | `top_p` | `max_completion_tokens`                   |
+| OpenRouter (chat) | `temperature` | `topP`  | `maxCompletionTokens`                     |
+| Ollama            | `temperature` | `top_p` | `num_predict` (nested in `options`)       |
+| BytePlus          | `temperature` | `top_p` | `max_tokens` (or `max_completion_tokens`) |
 
 `temperature` is the one key every provider names identically; token limits and
 some sampling options use provider-native names. Ollama nests all sampling under
@@ -324,20 +337,29 @@ runs.
 
 Current per-adapter status (#605):
 
-| Adapter                                      | Returns                                                                                           |
-| -------------------------------------------- | ------------------------------------------------------------------------------------------------- |
-| `openaiText` / `openaiChatCompletions`       | `true` (all supported models)                                                                     |
-| `anthropicText`                              | `true` for Claude 4.5+ (gated by `ANTHROPIC_COMBINED_TOOLS_AND_SCHEMA_MODELS`), `false` otherwise |
-| `geminiText`                                 | `true` for Gemini 3.x (gated by `GEMINI_COMBINED_TOOLS_AND_SCHEMA_MODELS`), `false` otherwise     |
-| `grokText`                                   | `true` for Grok 4 family (gated by `GROK_COMBINED_TOOLS_AND_SCHEMA_MODELS`), `false` otherwise    |
-| `groqText`                                   | `false` (Groq API rejects schema + tools + stream)                                                |
-| `openRouterText` / `openRouterResponsesText` | `false` (per-call resolution is a follow-up)                                                      |
-| `ollamaText`                                 | `false` (constrained-decoding vs tool-call grammar conflict)                                      |
+| Adapter                                      | Returns                                                                                               |
+| -------------------------------------------- | ----------------------------------------------------------------------------------------------------- |
+| `openaiText` / `openaiChatCompletions`       | `true` (all supported models)                                                                         |
+| `anthropicText`                              | `true` for Claude 4.5+ (gated by `ANTHROPIC_COMBINED_TOOLS_AND_SCHEMA_MODELS`), `false` otherwise     |
+| `geminiText`                                 | `true` for Gemini 3.x (gated by `GEMINI_COMBINED_TOOLS_AND_SCHEMA_MODELS`), `false` otherwise         |
+| `grokText`                                   | `true` for Grok 4 family (gated by `GROK_COMBINED_TOOLS_AND_SCHEMA_MODELS`), `false` otherwise        |
+| `groqText`                                   | `false` (Groq API rejects schema + tools + stream)                                                    |
+| `openRouterText` / `openRouterResponsesText` | `false` (per-call resolution is a follow-up)                                                          |
+| `ollamaText`                                 | `false` (constrained-decoding vs tool-call grammar conflict)                                          |
+| `byteplusText`                               | Per model — `true` only for the 10 ids in `BYTEPLUS_STRUCTURED_OUTPUT_CHAT_MODELS`, `false` otherwise |
 
 Subclasses can override to narrow the capability. When extending an
 adapter for a custom model that doesn't support the combination, return
 `false` explicitly.
 
+> **BytePlus has no JSON-mode fallback.** On the 8 chat models outside
+> `BYTEPLUS_STRUCTURED_OUTPUT_CHAT_MODELS`, Ark rejects `json_schema` _and_
+> `json_object`, so `false` here does not buy a degraded path — it only keeps
+> `response_format` out of the streaming chat request. `structuredOutput()`
+> throws and `structuredOutputStream()` emits `RUN_ERROR` on those models.
+> Note `seed-2-0-lite-260428` (the obvious default) is one of them; use
+> `seed-2-0-lite-260228` or `dola-seed-2-1-turbo-260628` for typed output.
+
 ### 6. OpenAI-Compatible Providers
 
 Any provider that implements the OpenAI **Chat Completions** API (DeepSeek,
@@ -443,6 +465,7 @@ Detailed per-adapter reference files:
 - [Grok Adapter](references/grok-adapter.md)
 - [Groq Adapter](references/groq-adapter.md)
 - [OpenRouter Adapter](references/openrouter-adapter.md)
+- [BytePlus Adapter](references/byteplus-adapter.md)
 
 ## Tension
 

package/skills/ai-core/adapter-configuration/references/anthropic-adapter.md

@@ -67,9 +67,7 @@ chat({
     // Tool choice
     tool_choice: { type: 'auto' },
     // Context management
-    context_management: {
-      /* BetaContextManagementConfig */
-    },
+    context_management: {/* BetaContextManagementConfig */},
     // MCP servers (max 20)
     mcp_servers: [
       {

package/skills/ai-core/adapter-configuration/references/byteplus-adapter.md

@@ -0,0 +1,148 @@
+# BytePlus Adapter Reference
+
+## Package
+
+```
+@tanstack/ai-byteplus
+```
+
+## Adapter Factories
+
+| Factory                 | Type          | Description                |
+| ----------------------- | ------------- | -------------------------- |
+| `byteplusText`          | Text/Chat     | ModelArk chat completions  |
+| `byteplusVideo`         | Video         | Seedance async task API    |
+| `byteplusImage`         | Image         | Seedream image generation  |
+| `byteplusSpeech`        | TTS           | Seed Speech text-to-speech |
+| `byteplusTranscription` | Transcription | Seed Speech ASR            |
+
+Each has a `createBytePlus*(model, apiKey, config?)` sibling that takes an
+explicit key instead of reading the environment.
+
+## Import
+
+```typescript
+import {
+  byteplusText,
+  byteplusVideo,
+  byteplusImage,
+} from '@tanstack/ai-byteplus'
+import { byteplusSpeech, byteplusTranscription } from '@tanstack/ai-byteplus'
+```
+
+## Key Chat Models
+
+| Model                        | Context | Notes                                                              |
+| ---------------------------- | ------- | ------------------------------------------------------------------ |
+| `dola-seed-2-1-turbo-260628` | 256K    | Flagship; text/image/video in; structured output; thinking summary |
+| `seed-2-0-lite-260428`       | 256K    | Good default; audio in; **no** structured output                   |
+| `seed-2-0-mini-260428`       | 256K    | Smallest 2.0; audio in; **no** structured output                   |
+| `seed-2-0-pro-260328`        | 256K    | Structured output; thinking summary                                |
+| `seed-2-0-lite-260228`       | 256K    | Structured output — use instead of `-260428` for typed output      |
+| `seed-1-6-flash-250715`      | 256K    | Fast/cheap; structured output                                      |
+| `glm-5-2-260617`             | 1024K   | Structured output; `reasoning_effort` incl. `none`/`xhigh`/`max`   |
+| `deepseek-v4-pro-260425`     | 1024K   | 384K output; no structured output                                  |
+| `deepseek-v3-2-251201`       | 128K    | Reasoning defaults **off**; no structured output                   |
+| `gpt-oss-120b-250805`        | 128K    | Only model accepting `thinking: { type: 'auto' }`                  |
+
+`BYTEPLUS_CHAT_MODELS` is the full list of 18 (also
+`seed-2-0-mini-260215`, `seed-2-0-code-preview-260328`, `seed-1-8-251228`,
+`seed-1-6-250915`, `seed-1-6-250615`, `seed-1-6-flash-250615`,
+`glm-4-7-251222`, `deepseek-v4-flash-260425`).
+
+Media models: `BYTEPLUS_VIDEO_MODELS` (Seedance —
+`dreamina-seedance-2-0-260128` / `-fast-260128` / `-mini-260615`,
+`seedance-1-5-pro-251215`, `seedance-1-0-pro-250528`,
+`seedance-1-0-pro-fast-251015`), `BYTEPLUS_IMAGE_MODELS` (Seedream —
+`dola-seedream-5-0-pro-260628`, `seedream-5-0-260128`,
+`seedream-5-0-lite-260128`, `seedream-4-5-251128`, `seedream-4-0-250828`),
+`BYTEPLUS_TTS_MODELS` (`seed-audio-1.0`) and
+`BYTEPLUS_TRANSCRIPTION_MODELS` (`seed-asr`).
+
+## Provider-Specific modelOptions
+
+```typescript
+chat({
+  adapter: byteplusText('dola-seed-2-1-turbo-260628'),
+  messages,
+  modelOptions: {
+    // Ark-only
+    thinking: { type: 'enabled' }, // 'enabled' | 'disabled' | 'auto' ('auto': gpt-oss-120b only)
+    reasoning_effort: 'medium', // 'none' | 'minimal' | 'low' | 'medium' | 'high' | 'xhigh' | 'max'
+    repetition_penalty: 1.1,
+    service_tier: 'default', // 'default' | 'flex' (flex = cheaper offline batch queue)
+    // Sampling (OpenAI-compatible names)
+    temperature: 0.7,
+    top_p: 0.9,
+    top_k: 40,
+    frequency_penalty: 0.5,
+    presence_penalty: 0.5,
+    seed: 42,
+    stop: ['\n\n'],
+    // Token limits — mutually exclusive
+    max_tokens: 2048,
+    // max_completion_tokens: 2048,
+    // Tool calling
+    tool_choice: 'auto',
+    parallel_tool_calls: true,
+    // Logging / attribution
+    logprobs: true,
+    top_logprobs: 5,
+    user: 'user-123',
+  },
+})
+```
+
+`response_format` is deliberately **not** a provider option — the chat
+activity owns it via `outputSchema`.
+
+## Environment Variables
+
+```
+ARK_API_KEY              # chat, video, image (falls back to BYTEPLUS_API_KEY)
+BYTEPLUS_VOICE_API_KEY   # TTS + transcription — a DIFFERENT product key
+```
+
+## Gotchas
+
+- **Two products, two keys.** ModelArk uses `Authorization: Bearer $ARK_API_KEY`
+  at `ark.ap-southeast.bytepluses.com`; Seed Speech uses
+  `X-Api-Key: $BYTEPLUS_VOICE_API_KEY` at `voice.ap-southeast-1.bytepluses.com`.
+  An Ark key on the voice host fails with `45000010 Invalid X-Api-Key`.
+- **Ark keys are region-isolated.** The default base URL is ap-southeast; the EU
+  endpoint serves chat and image only (no Seedance). Override with
+  `config.baseURL`.
+- **Structured output is per model and fails loud.** Only the 10 ids in
+  `BYTEPLUS_STRUCTURED_OUTPUT_CHAT_MODELS` accept `json_schema`; the other 8
+  reject `json_object` too, so there is no JSON-mode fallback. `glm-4-7-251222`
+  is excluded deliberately — it accepts the schema with `200` and then answers
+  in prose. `seed-2-0-lite-260428` is _not_ on the list.
+- **`reasoning_effort` + `thinking: { type: 'disabled' }` is a 400.** So is
+  sending both `max_tokens` and `max_completion_tokens`.
+- **`reasoning_effort` values are gated per model.** `minimal` / `low` /
+  `medium` / `high` are general; `none` and `xhigh` are accepted only by
+  `glm-5-2-260617`; `max` only by `glm-5-2-260617` and the two
+  `deepseek-v4-*-260425` models. `thinking: { type: 'auto' }` is
+  `gpt-oss-120b-250805` only.
+- **Reasoning is on by default** on every model except `deepseek-v3-2-251201`.
+- **`encrypted_content` round-trip.** The four thinking-summary models
+  (`dola-seed-2-1-turbo-260628`, `seed-2-0-lite-260428`,
+  `seed-2-0-mini-260428`, `seed-2-0-pro-260328`) emit an opaque signature over
+  the reasoning trace that BytePlus asks for back on the next turn. The adapter
+  round-trips it automatically over the same seam Anthropic thinking signatures
+  use: it is captured onto `STEP_FINISHED.signature`, which lands on the
+  thinking part as `ModelMessage.thinking[].signature` and is echoed back on the
+  following request. Code that persists and replays history by hand must **keep
+  the thinking part's `signature`** — not "message metadata", which is a
+  different field and carries nothing here. Losing it costs a reasoning-cache
+  hit, not the request: a live probe confirmed Ark accepts a turn whose
+  assistant message omits `encrypted_content`. A structured-output turn doesn't
+  capture the blob at all, for the same reason.
+- **Model ids are dated and retired aggressively.** Only ids verified against a
+  live request are exported; the published BytePlus lists include dead ones.
+  Bare ids mostly don't resolve — the `dola-` prefix is required on
+  `dola-seed-2-1-turbo-260628` and the `dreamina-` prefix on the Seedance 2.0
+  family, but adding `dola-` to an older id 404s.
+- **Media option applicability is per model and enforced server-side** — Ark
+  400s on an inapplicable Seedance field instead of ignoring it. See the
+  media-generation skill.

package/skills/ai-core/adapter-configuration/references/gemini-adapter.md

@@ -59,14 +59,10 @@ chat({
       },
     ],
     // Tool config
-    toolConfig: {
-      /* ToolConfig */
-    },
+    toolConfig: {/* ToolConfig */},
     // Structured output
     responseMimeType: 'application/json',
-    responseSchema: {
-      /* Schema */
-    },
+    responseSchema: {/* Schema */},
     // Cached content
     cachedContent: 'cachedContents/abc123',
     // Response modalities

package/skills/ai-core/adapter-configuration/references/groq-adapter.md

@@ -49,9 +49,7 @@ chat({
     // Response format
     response_format: {
       type: 'json_schema',
-      json_schema: {
-        /* ... */
-      },
+      json_schema: {/* ... */},
     },
     // Sampling
     temperature: 0.7,
@@ -71,9 +69,7 @@ chat({
     // Documents for context
     documents: [{ text: '...' }],
     // Search settings (for web search tool)
-    search_settings: {
-      /* SearchSettings */
-    },
+    search_settings: {/* SearchSettings */},
     // Service tier
     service_tier: 'auto', // 'auto' | 'on_demand' | 'flex' | 'performance'
     // Metadata

package/skills/ai-core/adapter-configuration/references/openai-adapter.md

@@ -64,9 +64,7 @@ chat({
     parallel_tool_calls: true,
     tool_choice: 'auto', // 'auto' | 'none' | 'required'
     // Structured output
-    text: {
-      /* ResponseTextConfig */
-    },
+    text: {/* ResponseTextConfig */},
     // Metadata (max 16 key-value pairs)
     metadata: { session_id: 'abc' },
     // Streaming

package/skills/ai-core/ag-ui-protocol/SKILL.md

@@ -8,7 +8,7 @@ description: >
   NDJSON format. For backends serving AG-UI events without client packages.
 type: sub-skill
 library: tanstack-ai
-library_version: '0.10.0'
+library_version: '0.42.0'
 sources:
   - 'TanStack/ai:docs/protocol/chunk-definitions.md'
   - 'TanStack/ai:docs/protocol/sse-protocol.md'