ChatMiddlewareContext

Defined in: packages/ai/src/activities/chat/middleware/types.ts:192

Stable context object passed to all middleware hooks. Created once per chat() invocation and shared across all hooks.

Type Parameters#

TContext#

TContext = unknown

Properties#

abort#

abort: (reason?) => void;

Defined in: packages/ai/src/activities/chat/middleware/types.ts:228

Abort the chat run with a reason

Parameters#

reason?#

string

Returns#

void

accumulatedContent#

accumulatedContent: string;

Defined in: packages/ai/src/activities/chat/middleware/types.ts:290

Accumulated text content for the current iteration

activities?#

optional activities?: readonly ActivityRecord[];

Defined in: packages/ai/src/activities/chat/middleware/types.ts:301

Frontend-only AG-UI activity sidecar. Never model input. Persistence writes this when an ActivityStore is configured. Optional so code that builds its own context keeps working; the engine always sets it.

activity#

activity: "chat";

Defined in: packages/ai/src/activities/chat/middleware/types.ts:257

Which activity this context describes — always 'chat'. Present so the chat context structurally satisfies the base GenerationMiddlewareContext, letting an observe-only middleware authored against the base (e.g. otelMiddleware) run on both chat and media activities.

capabilities#

capabilities: CapabilityRegistry;

Defined in: packages/ai/src/activities/chat/middleware/types.ts:310

Capability bookkeeping for this request. Populated by middleware setup hooks (via provide accessors) and read by later middleware (via get accessors). Prefer the accessors returned by createCapability over using this directly. Orthogonal to context (the user runtime context).

chunkIndex#

chunkIndex: number;

Defined in: packages/ai/src/activities/chat/middleware/types.ts:224

Running count of chunks yielded so far

context#

context: TContext;

Defined in: packages/ai/src/activities/chat/middleware/types.ts:241

Runtime context provided by chat() options

conversationId?#

optional conversationId?: string;

Defined in: packages/ai/src/activities/chat/middleware/types.ts:218

Deprecated#

Use threadId instead. Retained as an alias of threadId so middleware written before the AG-UI rename keeps working unchanged. Will be removed in a future major release.

createId#

createId: (prefix) => string;

Defined in: packages/ai/src/activities/chat/middleware/types.ts:303

Generate a unique ID with the given prefix

Parameters#

prefix#

string

Returns#

string

currentMessageId#

currentMessageId: string | null;

Defined in: packages/ai/src/activities/chat/middleware/types.ts:288

Current assistant message ID (changes per iteration)

defer#

defer: (promise) => void;

Defined in: packages/ai/src/activities/chat/middleware/types.ts:247

Defer a non-blocking side-effect promise. Deferred promises do not block streaming and are awaited after the terminal hook (onFinish/onAbort/onError).

Parameters#

promise#

Promise<unknown>

Returns#

void

emitCustomEvent#

emitCustomEvent: (name, value, options?) => void;

Defined in: packages/ai/src/activities/chat/middleware/types.ts:235

Push a CUSTOM chunk onto the chat stream immediately. The engine yields it as soon as it can (including while onConfig is still awaiting work such as a summarize call). Durability then flushes the event on its own, unless you pass { batch: true }.

Parameters#

name#

string

value#

Record<string, any>

options?#

EmitCustomEventOptions

Returns#

void

get#

get: <TValue>(capability) => TValue;

Defined in: packages/ai/src/activities/chat/middleware/types.ts:315

Read a provided capability by its handle. Equivalent to the handle's own get accessor (getX(ctx)); throws if the capability was never provided.

Type Parameters#

TValue#

TValue

Parameters#

capability#

Capability<TValue>

Returns#

TValue

getOptional#

getOptional: <TValue>(capability) => TValue | undefined;

Defined in: packages/ai/src/activities/chat/middleware/types.ts:320

Read a capability by its handle, returning undefined if it was never provided (never throws).

Type Parameters#

TValue#

TValue

Parameters#

capability#

Capability<TValue>

Returns#

TValue | undefined

hasTools#

hasTools: boolean;

Defined in: packages/ai/src/activities/chat/middleware/types.ts:283

Whether tools are configured

iteration#

iteration: number;

Defined in: packages/ai/src/activities/chat/middleware/types.ts:222

Current agent loop iteration (0-indexed)

messageCount#

messageCount: number;

Defined in: packages/ai/src/activities/chat/middleware/types.ts:281

Number of messages at the start of the request

messages#

messages: readonly ModelMessage<
  | string
  | ContentPart<unknown, unknown, unknown, unknown, unknown>[]
  | null>[];

Defined in: packages/ai/src/activities/chat/middleware/types.ts:295

Current messages array (read-only view)

model#

model: string;

Defined in: packages/ai/src/activities/chat/middleware/types.ts:261

Model identifier (e.g., 'gpt-5.5')

modelOptions?#

optional modelOptions?: Record<string, unknown>;

Defined in: packages/ai/src/activities/chat/middleware/types.ts:276

Provider-specific model options

options?#

optional options?: Record<string, unknown>;

Defined in: packages/ai/src/activities/chat/middleware/types.ts:274

Flattened generation options (metadata)

parentRunId?#

optional parentRunId?: string;

Defined in: packages/ai/src/activities/chat/middleware/types.ts:200

Interrupted or parent run correlated with this continuation.

phase#

phase: ChatMiddlewarePhase;

Defined in: packages/ai/src/activities/chat/middleware/types.ts:220

Current lifecycle phase

provide#

provide: <TValue>(capability, value) => void;

Defined in: packages/ai/src/activities/chat/middleware/types.ts:325

Provide a capability value. Equivalent to the handle's own provide accessor (provideX(ctx, value)). Typically called from setup.

Type Parameters#

TValue#

TValue

Parameters#

capability#

Capability<TValue>

value#

TValue

Returns#

void

provider#

provider: string;

Defined in: packages/ai/src/activities/chat/middleware/types.ts:259

Provider name (e.g., 'openai', 'anthropic')

requestId#

requestId: string;

Defined in: packages/ai/src/activities/chat/middleware/types.ts:194

Unique identifier for this chat request

runId#

runId: string;

Defined in: packages/ai/src/activities/chat/middleware/types.ts:198

AG-UI run identifier for correlating client and server events

signal?#

optional signal?: AbortSignal;

Defined in: packages/ai/src/activities/chat/middleware/types.ts:226

Abort signal from the chat request

source#

source: "server" | "client";

Defined in: packages/ai/src/activities/chat/middleware/types.ts:263

Source of the chat invocation — always 'server' for server-side chat

streamId#

streamId: string;

Defined in: packages/ai/src/activities/chat/middleware/types.ts:196

Unique identifier for this stream

streaming#

streaming: boolean;

Defined in: packages/ai/src/activities/chat/middleware/types.ts:265

Whether the chat is streaming

subagentRunId?#

optional subagentRunId?: string;

Defined in: packages/ai/src/activities/chat/middleware/types.ts:205

Set when this run is a subagent. The id on the child's SUBAGENT_STARTED and on every chunk it streams. Absent on a top-level run.

systemPrompts#

systemPrompts: SystemPrompt[];

Defined in: packages/ai/src/activities/chat/middleware/types.ts:270

System prompts configured for this chat

threadId#

threadId: string;

Defined in: packages/ai/src/activities/chat/middleware/types.ts:212

AG-UI thread identifier — a stable per-conversation ID used to correlate client and server devtools events. Resolves to the caller-provided threadId (or legacy conversationId), or an auto-generated value when neither is supplied.

toolNames?#

optional toolNames?: string[];

Defined in: packages/ai/src/activities/chat/middleware/types.ts:272

Names of configured tools, if any