Repairs TypeScript check-types across the monorepo and adds a CI gate so
regressions are caught going forward:
- core: bundler module resolution and strict-mode fixes
- sdk-js: bundler module resolution; keep codegen, formatter, packaging working
- react-core: fixes across components, hooks, and tests
- react-native: restore catch binding referenced by TypeError cause
- runtime: repair check-types and bound AI SDK schema inference
- web-inspector: nodenext import extensions, export Anchor
- remaining packages and node example: assorted check-types repairs
- deps: add missing type-only devDependencies
- license context driven from /info licenseStatus
- ci: run check-types in the static quality workflow
Squashed from 12 commits for a single, easily-revertable change.
Adds the AG-UI standard interrupt flow (RUN_FINISHED outcome:interrupt + resume array) alongside the legacy on_interrupt path.
- core: forward the standard resume array through runAgent.
- react-core / vue / react-native: useInterrupt handles standard interrupts with resolve()/cancel(), surfaces the primary + full interrupt set, and persists each resolved tool-backed interrupt as a tool-result message so multi-turn conversations stay well-formed (no dangling tool call -> no tool-call loop).
- runtime BuiltInAgent: native interrupts for the aisdk + tanstack factory paths via each SDK's needsApproval primitive (tool-approval-request / CUSTOM approval-requested -> outcome:interrupt); classic interrupt-tool emission + ctx.interrupt() factory primitive; idempotent resume injection mapped to each SDK's native tool-result; getCapabilities advertises humanInTheLoop.interrupts.
- docs: document standard interrupt support.
Verified across core/react-core/runtime unit suites and a real-model multi-turn run on both aisdk and tanstack.
The TanStack stream converter stopped at the first per-turn RUN_FINISHED,
assuming tools are executed client-side. That truncated runs whose tools are
executed by chat() itself (MCP servers, provider tools): the TOOL_CALL_RESULT
and the model's final answer were dropped, so MCP-backed turns returned nothing.
- convertTanStackStream: drop TanStack's per-turn RUN_STARTED/RUN_FINISHED (the
Agent wrapper owns the outer pair) and convert every turn's events; dedupe
tool START/END by id; surface RUN_ERROR instead of silently dropping it.
- convertInputToTanStackAI: return input.tools as TanStack client-side tools so
the frontend's generative-UI / HITL tools work in factory mode, and sanitize
their JSON Schema (close open objects) so OpenAI accepts them.
Adds converter + input tests (multi-turn, dedup, error surfacing, client-tool
conversion, schema sanitizing).
A single unavailable MCP server (down, 5xx, timeout, bad auth) no longer
fails the whole run - it is skipped with an error log and the run
continues with healthy servers and the agent's own tools.
Move enterprise-learning MCP attachment out of the BuiltInAgent-specific
path and the intelligence run handler into a single request-scoped hook:
- `attachIntelligenceEnterpriseLearning` (agent-utils) attaches
`@ag-ui/mcp-middleware` via `configureAgentForRequest`, gated on
`ɵisEnterpriseLearningEnabled()`, resolving the user via `identifyUser`
and the project apiKey.
- Called from `handleRunAgent`; the old `forwardedProps.auth` MCP plumbing
in `intelligence/run.ts` and the BuiltInAgent attach in `agent/index.ts`
are removed.
- Add released `@ag-ui/mcp-middleware@0.0.1` dependency (lockfile +
`@ag-ui/client` override). Drops the obsolete intelligence-mcp-helper test.
Two CR comments addressed:
- Rename the local destructure of forwardedProps.auth.copilotkitIntelligence
from 'cki' to 'cpki' so it matches the project-wide abbreviation already
used in metadata fields (cpki_event_id, cpki_event_seq, etc).
- Replace the inline 'X-Cpki-User-Id' string literal with the existing
INTELLIGENCE_USER_ID_HEADER constant exported from intelligence-platform/client.
Applies to the runtime auto-attach in agent/index.ts and to the three
test sites in intelligence-mcp-helper.test.ts so the user-side and
runtime-side stay in sync.
Move the per-request Intelligence MCP bag from
forwardedProps.copilotkitIntelligence to
forwardedProps.auth.copilotkitIntelligence so the Intelligence-side
redaction policy strips it. The 'auth' namespace is the convention for
credentials; persistence sinks (Postgres, Redis, S3) and FE replay
paths in apps/realtime-gateway already strip everything under it.
Updates:
- Emitter (handlers/intelligence/run.ts): merge the bag into a single
forwardedProps.auth object alongside any upstream auth keys, and
only emit the auth namespace when there is something to put in it.
- Reader (agent/index.ts): read from forwardedProps.auth.copilotkitIntelligence
instead of forwardedProps.copilotkitIntelligence.
- Tests (intelligence-mcp-helper.test.ts): three fixtures rewritten
to the nested shape.
Runtime side (intelligence/run.ts): when `runtime.intelligence.mcpServer`
is enabled, build a `copilotkitIntelligence` bag carrying the resolved
user-id, project apiKey, and the platform MCP URL, and pass it through
on `RunAgentInput.forwardedProps` to the agent. Skipped when the flag
is off — runs that don't go through this Intelligence path simply
don't see the bag.
Agent side (BuiltInAgent's run code): if `forwardedProps.copilotkitIntelligence`
contains all three string values AND the user's static `config.mcpServers`
doesn't already include the same URL, append a per-request
`MCPClientConfigHTTP`. Its `options.fetch` closes over apiKey + userId
and stamps `Authorization: Bearer <apiKey>` and `X-Cpki-User-Id:
<userId>` on every outbound MCP call. The custom fetch is the MCP
TypeScript SDK's documented extension point for per-request header
injection — no extra wrapper class, no separate framework concept.
The agent class is otherwise untouched: no new fields, no per-request
side channels, no typed reference to `CopilotKitIntelligence`. Other
agents that don't read `forwardedProps.copilotkitIntelligence` ignore
the bag.
Pulls in the AI SDK's stable `createMCPClient` export (rename from
`experimental_createMCPClient`) — the experimental name was deprecated;
`mcp-clients.test.ts`'s mock setup follows.
TanStack's chat() engine runs a multi-turn agent loop: after the model
returns tool calls, it tries to execute them via processToolCalls().
Frontend-only tools (like render_pie_chart) are unknown to TanStack, so
executeToolCalls() treats them as errors and buildToolResultChunks()
re-emits TOOL_CALL_END without a preceding TOOL_CALL_START. The ag-ui
verify middleware rejects this duplicate.
Fix: track a runFinished flag in convertTanStackStream and discard all
events after the first RUN_FINISHED, which marks the boundary between
the streaming pass and TanStack's internal tool execution loop.
Also adds built-in-agent to docker-compose.local.yml and local-ports.json.
## Summary
- Providers like `@ai-sdk/openai-compatible` emit sequential IDs
(`txt-0`, `reasoning-0`, `msg-0`) that are treated as unique but collide
across requests
- The existing check only caught the literal `"0"` — expanded to match
the `^(txt|reasoning|msg)-0$` pattern
- Both `text-start` and `reasoning-start` event handlers now use
`randomUUID()` for these non-unique IDs
## Test plan
- [x] Added `provider-id-collision.test.ts` with 4 tests covering txt-0,
reasoning-0, msg-0, and legitimate ID preservation
- [x] All 1226 runtime tests pass (including existing
config-tools-execution tests)
Fixes#3410, #3623
- Add per-agent error isolation in /info handler so a single
getCapabilities() failure doesn't 500 the whole endpoint
- Normalize getCapabilities() to return undefined (not {}) when no
capabilities are set, matching the sync getter
- Replace instanceof ProxiedCopilotRuntimeAgent with duck-type check
in useCapabilities hook for extensibility
- Add JSDoc warning about shallow-merge behavior on capabilities config
- Add useCapabilities hook tests (5 cases covering both branches)
- Add per-agent error isolation test for get-runtime-info
Providers like @ai-sdk/openai-compatible emit sequential IDs (txt-0,
reasoning-0, msg-0) that collide across requests. The existing check
only caught the literal "0". Expand the check to match the
^(txt|reasoning|msg)-0$ pattern and always use randomUUID() for these.
TanStackChatMessage.content uses any[] for multimodal so
messages are directly passable to any adapter without casts.
Add env.d.ts with vite/client types for CSS imports.
- Fix convertInputToTanStackAI silently dropping multimodal
content (images, audio, video, documents) by converting
AG-UI content parts to TanStack AI ContentPart format
- Switch react-router example from Hono server to a React
Router resource route using createCopilotRuntimeHandler
- Both agents now use BuiltInAgent factory pattern
- Remove hono/react-router-hono-server deps from example
- Fix all doc imports to use @copilotkit/runtime/v2 (symbols only
exported from v2 entry point)
- Replace v1 copilotRuntimeNextJSAppRouterEndpoint with v2 pattern
(CopilotRuntime + createCopilotEndpoint + InMemoryAgentRunner)
- Add warning callout on TanStack reasoning example (reasoning events
not surfaced by TanStack converter)
- Fix randomUUID imports in converters: use @copilotkit/shared instead
of Node.js crypto to match codebase convention
- Fix collectEvents timer leak in test-helpers.ts (add clearTimeout)
- Fix JSDoc/comments: convertInputToTanStackAI accurately describes
role allowlist, TanStackInputResult.messages comment updated
- Close TOCTOU window: set abortController synchronously before Observable
creation in classic run(), matching factory run() pattern (M3)
- Add concurrent run guard to classic run() (C1)
- Add threadId/runId to RUN_ERROR events in classic mode (I2)
- Check both 'output' and 'result' property names in classic tool-result (C2/M2)
- Add try/catch around JSON.stringify in classic tool-result (I4/M2)
- Add undefined guards before emitting state snapshot/delta events (I3)
- Guard assignToolsToAgents against factory-mode agent configs (M1)
- Improve error handling in classic error case with proper fallback (C3)
- Improve TanStack converter TODO comment with known gap details (M5)
- Document @ts-expect-error on clone() middlewares access (S1)
- Add JSDoc to AgentFactoryContext.abortController (M4)
- Fix duplicate import and clean up react-router example
- Update docs AgentFactoryContext reference
- Add converter imports (convertAISDKStream, convertTanStackStream) to agent/index.ts
- Update canOverride() to return false for factory configs
- Add factory-path early return in run() that delegates to runFactory()
- Add private runFactory() method that handles aisdk/tanstack/custom factory configs
with full lifecycle management (RUN_STARTED, RUN_FINISHED, RUN_ERROR, abort support)
- Fix copilot-runtime.ts: change BuiltInAgentConfiguration["tools"] indexed access
types to BuiltInAgentClassicConfig["tools"] since tools only exist on classic configs
Source fixes:
- Fix TOCTOU race in concurrent run guard — set abortController
synchronously before Observable creation, closing the window where
two run() calls could both pass the guard
- Add JSON.stringify try/catch in TanStack TOOL_CALL_RESULT handler
(matching existing aisdk.ts protection against circular refs)
- Use role allowlist (user/assistant/tool) instead of denylist in
convertInputToTanStackAI to exclude activity/reasoning roles
- Change error+abort handler from break to return in aisdk converter
to stop processing after error during abort
- Fix version-pinned comments (remove "AI SDK 5.0" reference)
Test fixes:
- Add timedOut flag to collectEventsIncludingErrors to distinguish
hung observables from normal completion
- Add clearTimeout on success/error to prevent timer leaks
- Remove timing-dependent setTimeout in concurrent run guard test
Docs fixes:
- Split imports: copilotRuntimeNextJSAppRouterEndpoint from
@copilotkit/runtime, Agent/converters from @copilotkit/runtime/v2
- Fix forwardedProps example: use CopilotKit properties prop
- Fix useCopilotChat → useCoAgent for state management reference
- Fix model syntax colon → slash (openai/gpt-4o)
Critical fixes:
- Guard concurrent run() calls — throw if agent is already running
- Check both "output" and "result" properties on AI SDK tool-result events
for compatibility with older AI SDK versions
- Handle undefined/missing error in AI SDK error events with proper
Error wrapping instead of throwing undefined
Important fixes:
- Add TOOL_CALL_RESULT handling to TanStack converter (was silently dropped)
- Include threadId and runId in RUN_ERROR events for run correlation
- Guard state tool events (STATE_SNAPSHOT/STATE_DELTA) against undefined
snapshot/delta fields before emitting
- Wrap JSON.stringify in tool-result with try/catch to handle circular refs
- Fix collectEventsIncludingErrors to distinguish error vs normal completion
Test coverage:
- Add tests for convertInputToTanStackAI (message filtering, tool call
mapping, context injection, state serialization)
- Add tests for state management tool results in AI SDK converter
- Add tests for async factory (Promise-returning) paths
- Add tests for tool-result property compatibility (output vs result)
- Add tests for error event handling edge cases
- Add tests for concurrent run guard and RUN_ERROR correlation fields
- Replace flaky setTimeout-based abort test with signal-based sync
Other:
- Add exhaustiveness check (default: never) in config type switch
- Document clone() @ts-expect-error coupling with AbstractAgent.middlewares
- Fix docs: remove incorrect "v6" version reference, fix wrong hook name
Nathan 🔶 Tarbert