@backendkit-labs/agent-web 0.6.5 → 0.6.6

package/README.md

@@ -1,613 +1,613 @@
-# @backendkit-labs/agent-web
-
-HTTP/WebSocket server layer for [@backendkit-labs/agent-core](https://www.npmjs.com/package/@backendkit-labs/agent-core).
-
-Wraps `AgentEngine` in a Fastify 5 server with streaming WebSocket sessions, JWT authentication, Prometheus metrics, rate limiting, a real-time monitor bus, and a flexible trigger system (cron jobs, webhooks, BullMQ queues).
-
-[![npm](https://img.shields.io/npm/v/@backendkit-labs/agent-web)](https://www.npmjs.com/package/@backendkit-labs/agent-web)
-
-## Table of contents
-
-- [Installation](#installation)
-- [Quick start](#quick-start)
-- [AgentServer options](#agentserver-options)
-- [HTTP and WebSocket routes](#http-and-websocket-routes)
-- [WebSocket protocol](#websocket-protocol)
-- [EngineFactory pattern](#enginefactory-pattern)
-- [SessionManager](#sessionmanager)
-- [JWT authentication](#jwt-authentication)
-- [Prometheus metrics](#prometheus-metrics)
-- [Rate limiting](#rate-limiting)
-- [Session metadata store](#session-metadata-store)
-- [MonitorBus](#monitorbus)
-- [TriggerBus — cron, webhooks, queues](#triggerbus--cron-webhooks-queues)
-- [HeadlessTransport](#headlesstransport)
-- [Full production example](#full-production-example)
-
----
-
-## Installation
-
-```bash
-npm install @backendkit-labs/agent-web @backendkit-labs/agent-core
-```
-
-Optional integrations (install only what you need):
-
-```bash
-npm install jsonwebtoken          # JWT auth
-npm install prom-client           # Prometheus metrics
-npm install @fastify/rate-limit   # rate limiting
-npm install node-cron             # cron triggers
-npm install bullmq                # queue triggers
-npm install ioredis               # Redis session store
-```
-
----
-
-## Quick start
-
-```typescript
-import { AgentServer } from '@backendkit-labs/agent-web';
-import type { EngineFactory } from '@backendkit-labs/agent-web';
-import { AgentEngine, AgentRegistry, ToolRegistry, ProviderRegistry } from '@backendkit-labs/agent-core';
-
-const engineFactory: EngineFactory = (sessionId, transport) =>
-  new AgentEngine({
-    transport,
-    sessionId,
-    // ...your engine config
-  });
-
-const server = new AgentServer({
-  port:          3000,
-  engineFactory,
-  cors:          '*',
-});
-
-await server.start();
-// WS  /chat      — streaming chat sessions
-// WS  /monitor   — real-time read-only observer
-// GET /health    — { status, sessions, monitors }
-// GET /sessions  — active session list
-// DELETE /sessions/:id — close a session
-```
-
----
-
-## AgentServer options
-
-```typescript
-interface AgentServerOptions {
-  // Required
-  engineFactory: EngineFactory;       // creates a new AgentEngine per session
-
-  // Network
-  port?:   number;                    // default: 3000
-  host?:   string;                    // default: '127.0.0.1' (use '0.0.0.0', with auth, for remote callers)
-  cors?:   string | string[];         // default: '*' (none with strict)
-  strict?: boolean;                   // fail-closed defaults, see "Strict mode"
-  logger?: boolean | object;          // Fastify logger, default: false
-
-  // Auth
-  auth?: AuthHook;                    // called on every request before handlers
-
-  // Rate limiting (requires @fastify/rate-limit)
-  rateLimiting?: {
-    max?:        number;              // max requests per IP per window, default: 100
-    timeWindow?: string;             // e.g. '1 minute', '30 seconds'
-  };
-
-  // Prometheus metrics (requires prom-client)
-  metrics?: {
-    enabled:          boolean;        // must be true to activate
-    path?:            string;         // default: '/metrics'
-    prefix?:          string;         // metric name prefix
-    collectNodeMetrics?: boolean;     // include Node.js process metrics
-  };
-
-  // Session management
-  maxIdleMs?:   number;               // idle timeout before session is closed
-  maxSessions?: number;               // max concurrent sessions
-  store?:       SessionMetaStore;     // durable metadata store (default: in-memory)
-
-  // Triggers
-  triggers?: TriggerBus;             // wires webhook routes and starts cron jobs
-}
-```
-
----
-
-## HTTP and WebSocket routes
-
-| Method | Path | Description |
-|--------|------|-------------|
-| `GET` | `/health` | Server health: `{ status, sessions, monitors, uptime }` |
-| `GET` | `/sessions` | Array of active session metadata |
-| `DELETE` | `/sessions/:id` | Force-close a session |
-| `GET` | `/metrics` | Prometheus metrics (if enabled) |
-| `WS` | `/chat` | Streaming chat — one session per connection |
-| `WS` | `/monitor` | Read-only observer of all sessions |
-| `POST` | `/webhooks/:name` | Webhook trigger (registered via TriggerBus) |
-
----
-
-## WebSocket protocol
-
-### `/chat` — chat sessions
-
-**Client → server:**
-
-```json
-{ "type": "input",  "content": "Your message here" }
-{ "type": "abort" }
-```
-
-**Server → client** (streamed `AgentEvent` frames):
-
-```json
-{ "type": "run_start",     "sessionId": "abc-123" }
-{ "type": "token",         "content": "Hello" }
-{ "type": "tool_call",     "name": "search", "args": { "query": "..." } }
-{ "type": "tool_result",   "name": "search", "result": "..." }
-{ "type": "agent_switch",  "from": "general", "to": "security-reviewer" }
-{ "type": "done",          "final": "Full assistant response" }
-{ "type": "error",         "message": "Something went wrong" }
-```
-
-### `/monitor` — real-time observer
-
-Connect to observe all sessions without sending any messages:
-
-```json
-{ "type": "monitor_hello",         "sessionIds": ["abc-123", "def-456"] }
-{ "type": "monitor_session_open",  "sessionId": "ghi-789" }
-{ "type": "monitor_session_close", "sessionId": "ghi-789" }
-{ "sessionId": "abc-123", "type": "run_input", "content": "..." }
-{ "sessionId": "abc-123", "type": "token",     "content": "..." }
-{ "sessionId": "abc-123", "type": "done",      "final": "..." }
-```
-
----
-
-## EngineFactory pattern
-
-`EngineFactory` is a function `(sessionId: string, transport: Transport) => AgentEngine`. Each incoming WebSocket connection gets its own isolated engine instance.
-
-### Minimal factory
-
-```typescript
-import type { EngineFactory } from '@backendkit-labs/agent-web';
-import { AgentEngine } from '@backendkit-labs/agent-core';
-
-const engineFactory: EngineFactory = (sessionId, transport) =>
-  new AgentEngine({
-    transport,
-    sessionId,
-    model:           { provider: 'openai', id: 'gpt-4o' },
-    agents,
-    tools,
-    providers,
-    defaultProvider: 'openai',
-    defaultAgentId:  'general',
-    maxIterations:   50,
-    iterationMode:   'auto',
-  });
-```
-
-### Factory with per-session memory
-
-```typescript
-import { createMemorySystem } from '@backendkit-labs/agent-core';
-import * as path from 'path';
-import * as os from 'os';
-
-const engineFactory: EngineFactory = (sessionId, transport) => {
-  // Isolated memory per session
-  const sessDir = path.join(os.homedir(), '.bk-agent', 'sessions', sessionId);
-  const memory = createMemorySystem({
-    episodic:  { provider: 'json', path: path.join(sessDir, 'episodes.json') },
-    semantic:  { provider: 'json', path: path.join(sessDir, 'knowledge.json') },
-    procedural:{ provider: 'json', path: path.join(sessDir, 'patterns.json') },
-  });
-
-  return new AgentEngine({
-    transport,
-    sessionId,
-    memorySystem: memory,
-    ...baseConfig,
-  });
-};
-```
-
-### Factory with per-user context
-
-Each WebSocket connection to `/chat` gets its own session with a random `sessionId`, and the factory builds a fresh engine for it. When `auth` is set, the authenticated user (the JWT `sub`) becomes the **owner** of the session: other users cannot see it in `/sessions` or `/monitor`. To load per-user settings in the engine, keep your own map from session to user, or read them inside your tools.
-
-```typescript
-const engineFactory: EngineFactory = (sessionId, transport) =>
-  new AgentEngine({
-    ...baseConfig,
-    transport,
-    sessionId,                 // also the session's id in metrics, monitor and audit
-  });
-```
-
----
-
-## SessionManager
-
-`SessionManager` creates, tracks and cleans up sessions. `AgentServer` creates one internally; you can also use it on your own WebSocket server.
-
-```typescript
-import { SessionManager } from '@backendkit-labs/agent-web';
-
-const sessions = new SessionManager({
-  engineFactory,
-  maxIdleMs:   30 * 60 * 1000,  // close after 30 min of inactivity (default)
-  maxSessions: 100,             // reject new sessions beyond this (default)
-  monitorBus:  myMonitorBus,
-  metrics:     myMetricsCollector,
-  store:       myMetaStore,
-});
-
-const session = sessions.create(socket, userId);   // one per WebSocket connection; userId is the owner
-sessions.get(session.id);
-sessions.list();                                   // [{ id, createdAt, lastActivityAt, userId }]
-sessions.destroy(session.id);
-sessions.close();                                  // all sessions, on shutdown
-```
-
----
-
-## JWT authentication
-
-`JwtAuth` validates Bearer tokens on HTTP routes and on the WebSocket handshake. Browsers cannot set headers on a WebSocket, so the token can also travel as the subprotocol `bearer, <jwt>`. A `?token=` query parameter still works unless `allowQueryToken: false`.
-
-### Strict mode
-
-For a server that other people reach, turn on both switches:
-
-```ts
-const auth = new JwtAuth({ secret: process.env.JWT_SECRET!, allowQueryToken: false }).hook();
-new AgentServer({ engineFactory, host: '0.0.0.0', auth, strict: true, cors: ['https://app.example.com'] });
-```
-
-- `allowQueryToken: false` ignores `?token=` (a token in a URL lands in logs, proxies and history). A browser sends the
-  token as the WebSocket subprotocol: `new WebSocket(url, ['bearer', jwt])`; `useAgentSession({ url, token })` in
-  `agent-web-ui` does it for you.
-- `strict: true` allows no CORS origin unless `cors` lists it, adds `X-Content-Type-Options`, `X-Frame-Options: DENY` and
-  `Referrer-Policy: no-referrer`, and refuses (throws) to listen on a non-loopback `host` without `auth`.
-
-### Setup
-
-```typescript
-import { AgentServer, JwtAuth } from '@backendkit-labs/agent-web';
-
-const jwt = new JwtAuth({
-  secret:     process.env.JWT_SECRET!,   // or publicKey for RS256 / ES256
-  algorithms: ['HS256'],                 // accepted algorithms
-  audience:   'agents',                  // optional checks
-  issuer:     'https://login.company.com',
-});
-
-const server = new AgentServer({ port: 3000, engineFactory, auth: jwt.hook() });
-```
-
-The token must carry `sub` (the user id, owner of the sessions it opens); `email` and `roles` are read when present. `adminRole` in `AgentServerOptions` names the role that may see every session.
-
-### Issuing tokens
-
-`JwtAuth` only verifies. Issue tokens in your login endpoint with any JWT library and the same secret or key pair:
-
-```typescript
-import jwt from 'jsonwebtoken';
-
-const token = jwt.sign({ sub: 'user-123', roles: ['admin'] }, process.env.JWT_SECRET!, { expiresIn: '24h' });
-// The client sends it as Authorization: Bearer <token>, or as the WebSocket subprotocol `bearer, <token>`.
-```
-
-### Custom auth hook
-
-If you use a different auth system (API keys, OAuth tokens), implement `AuthHook` directly:
-
-```typescript
-import type { AuthHook } from '@backendkit-labs/agent-web';
-
-const apiKeyAuth: AuthHook = async (req, reply) => {
-  const key = req.headers['x-api-key'];
-  if (!key || !validApiKeys.has(key as string)) {
-    reply.code(401).send({ error: 'Unauthorized' });
-  }
-};
-
-const server = new AgentServer({ auth: apiKeyAuth, ...opts });
-```
-
----
-
-## Prometheus metrics
-
-Requires `prom-client` to be installed. Exposes a `/metrics` endpoint with standard Prometheus counters and histograms.
-
-```typescript
-const server = new AgentServer({
-  ...opts,
-  metrics: {
-    enabled:            true,
-    path:               '/metrics',         // default
-    prefix:             'bk_agent_',        // prepended to all metric names
-    collectNodeMetrics: true,               // include Node.js process metrics
-  },
-});
-```
-
-### Exposed metrics
-
-| Metric | Type | Description |
-|--------|------|-------------|
-| `bk_agent_sessions_active` | Gauge | Currently open sessions |
-| `bk_agent_runs_total{status}` | Counter | Total runs by status (`success`, `error`) |
-| `bk_agent_run_duration_seconds` | Histogram | Run latency distribution |
-| Node.js metrics | Various | Memory, CPU, GC (when `collectNodeMetrics: true`) |
-
-### Using MetricsCollector directly
-
-```typescript
-import { MetricsCollector } from '@backendkit-labs/agent-web';
-
-const collector = new MetricsCollector({ prefix: 'my_agent_' });
-
-// Record a run
-const endRun = collector.runStarted();
-// ... after it completes:
-endRun('success');   // or endRun('error')
-
-collector.sessionOpened();      // and sessionClosed(), messageReceived()
-
-// Prometheus text format (serve it with collector.contentType)
-const text = await collector.collect();
-// # HELP my_agent_runs_total ...
-// my_agent_runs_total{status="success"} 42
-```
-
----
-
-## Rate limiting
-
-Requires `@fastify/rate-limit`. Applied globally to all routes:
-
-```typescript
-const server = new AgentServer({
-  ...opts,
-  rateLimiting: {
-    max:        60,          // max 60 requests per IP
-    timeWindow: '1 minute',  // sliding window
-  },
-});
-```
-
-When the limit is exceeded, the server responds with:
-```json
-{ "statusCode": 429, "error": "Too Many Requests", "message": "Rate limit exceeded" }
-```
-
----
-
-## Session metadata store
-
-By default, session metadata lives in memory and is lost on restart. Use `RedisSessionMetaStore` for durable metadata across restarts and horizontal scaling.
-
-### In-memory (default)
-
-```typescript
-import { InMemorySessionMetaStore } from '@backendkit-labs/agent-web';
-
-const server = new AgentServer({
-  ...opts,
-  store: new InMemorySessionMetaStore(),
-});
-```
-
-### Redis store (requires ioredis or node-redis)
-
-`RedisSessionMetaStore` is duck-typed — it accepts any Redis client that implements `.get()`, `.set()`, and `.del()`:
-
-```typescript
-import { RedisSessionMetaStore } from '@backendkit-labs/agent-web';
-import Redis from 'ioredis';
-
-const redis = new Redis(process.env.REDIS_URL!);
-
-const server = new AgentServer({
-  ...opts,
-  store: new RedisSessionMetaStore(redis, {
-    ttlSeconds: 86400,    // keep session metadata for 24h
-    prefix:     'bk:sess:', // key prefix
-  }),
-});
-```
-
----
-
-## MonitorBus
-
-`MonitorBus` broadcasts all session events to connected `/monitor` WebSocket observers. Used by the Kanban board in `@backendkit-labs/agent-web-ui`.
-
-### Standalone usage
-
-```typescript
-import { MonitorBus } from '@backendkit-labs/agent-web';
-
-const monitor = new MonitorBus();
-
-// Subscribe to all events (e.g. for logging or alerting)
-const unsub = monitor.subscribe((frame) => {
-  if (frame.type === 'done') {
-    console.log(`[${frame.sessionId}] run completed`);
-  }
-  if (frame.type === 'error') {
-    alerting.notify(`Agent error in session ${frame.sessionId}: ${frame.message}`);
-  }
-});
-
-// Clean up
-unsub();
-```
-
----
-
-## TriggerBus — cron, webhooks, queues
-
-`TriggerBus` runs agents without a WebSocket client: on a schedule, on an incoming webhook, or for each job of a BullMQ queue. Each run gets a fresh engine from the factory and a `HeadlessTransport`, and shows up in the monitor.
-
-Every trigger has the same three callbacks:
-
-- `buildPrompt(payload)` returns the prompt. `payload` is `undefined` for cron, the parsed body for a webhook, and `job.data` for a queue job.
-- `onComplete(result)` receives `{ sessionId, triggerName, firedAt, durationMs, output, events }`.
-- `onError(err, triggerName)` is called when the run throws.
-
-### Cron trigger (requires `node-cron`)
-
-```typescript
-import { TriggerBus } from '@backendkit-labs/agent-web';
-
-const triggers = new TriggerBus(engineFactory)
-  .addCron({
-    name:        'daily-report',
-    schedule:    '0 8 * * 1-5',     // weekdays at 8 AM
-    buildPrompt: () => 'Generate the daily engineering standup summary from recent commits.',
-    runOnStart:  false,             // true fires once when the bus starts
-    onComplete:  ({ output }) => slack.send('#daily', output),
-    onError:     (err, name) => logger.error(`[${name}]`, err),
-  })
-  .addCron({
-    name:        'health-check',
-    schedule:    '*/15 * * * *',
-    buildPrompt: () => 'Check all API endpoints and report any that are down.',
-    onComplete:  ({ output }) => { if (output.includes('DOWN')) pagerduty.alert(output); },
-  });
-
-const server = new AgentServer({ ...opts, triggers });
-await server.start();   // cron jobs start with the server
-```
-
-### Webhook trigger
-
-```typescript
-const triggers = new TriggerBus(engineFactory)
-  .addWebhook({
-    name:        'github-pr',
-    path:        '/webhooks/github',          // POST /webhooks/github (method: 'GET' | 'POST' | 'PUT')
-    secret:      process.env.GITHUB_SECRET,   // HMAC X-Hub-Signature-256 over the raw body
-    buildPrompt: (body) => {
-      const pr = (body as { pull_request: { number: number; title: string; diff_url: string } }).pull_request;
-      return `Review PR #${pr.number}: "${pr.title}"\n${pr.diff_url}`;
-    },
-    onComplete:  ({ output }) => github.comment(output),
-  });
-```
-
-The webhook answers immediately and the run continues in the background; use `onComplete` to act on the result. If the payload is needed there, capture it in `buildPrompt`.
-
-```bash
-curl -X POST http://localhost:3000/webhooks/github \
-  -H "Content-Type: application/json" \
-  -H "X-Hub-Signature-256: sha256=<hmac of the exact body>" \
-  -d '{ "pull_request": { "number": 42, "title": "Add auth module", "diff_url": "..." } }'
-```
-
-### Queue trigger (BullMQ — requires `bullmq`)
-
-```typescript
-const triggers = new TriggerBus(engineFactory)
-  .addQueue({
-    name:        'document-processor',
-    queueName:   'documents',
-    connection:  process.env.REDIS_URL!,   // URL, or { host, port, password, username, db, tls }
-    concurrency: 3,                        // default 1
-    buildPrompt: (data) => `Summarize and extract key points:\n\n${(data as { content: string }).content}`,
-    onComplete:  ({ output }) => saveSummary(output),
-  });
-```
-
-A failed run is rethrown to BullMQ, so the job is marked failed and its retry policy applies. Jobs are published with BullMQ as usual (`new Queue('documents', { connection }).add('process', { id, content }, { attempts: 3 })`).
-
----
-
-## HeadlessTransport
-
-A `Transport` with no socket: it buffers every `AgentEvent` in memory. `TriggerBus` uses it for autonomous runs; use it to run an engine from a script.
-
-```typescript
-import { HeadlessTransport } from '@backendkit-labs/agent-web';
-
-const transport = new HeadlessTransport();
-const engine = new AgentEngine({ ...opts, transport });
-await engine.run('Generate a daily report');
-
-transport.getOutput();   // concatenated text of the 'token' events
-transport.getEvents();   // every event, in order
-```
-
----
-
-## Full production example
-
-JWT auth, Redis sessions, Prometheus metrics, a cron job, a webhook and a queue:
-
-```typescript
-import { AgentServer, JwtAuth, TriggerBus, RedisSessionMetaStore } from '@backendkit-labs/agent-web';
-import { createEnterpriseSetup } from '@backendkit-labs/agent-enterprise';
-import { ProviderRegistry } from '@backendkit-labs/agent-core';
-import { DeepSeekProvider } from '@backendkit-labs/agent-providers';
-import Redis from 'ioredis';
-
-const providers = new ProviderRegistry();
-providers.register('deepseek', new DeepSeekProvider({ apiKey: process.env.DEEPSEEK_API_KEY!, model: 'deepseek-chat' }));
-
-const enterprise = createEnterpriseSetup({ vaultPath: process.env.VAULT_PATH!, providers, defaultProvider: 'deepseek' });
-await enterprise.indexAll();
-
-const triggers = new TriggerBus(enterprise.engineFactory)
-  .addCron({
-    name:        'weekly-digest',
-    schedule:    '0 9 * * 1',
-    buildPrompt: () => 'Generate a weekly digest of all knowledge base changes.',
-    onComplete:  ({ output }) => slack.send('#general', output),
-  })
-  .addWebhook({
-    name:        'support-ticket',
-    path:        '/webhooks/support',
-    secret:      process.env.WEBHOOK_SECRET,
-    buildPrompt: (body) => {
-      const t = body as { id: string; subject: string; description: string };
-      return `Handle support ticket #${t.id}: ${t.subject}\n${t.description}`;
-    },
-  })
-  .addQueue({
-    name:        'email-processor',
-    queueName:   'emails',
-    connection:  process.env.REDIS_URL!,
-    concurrency: 5,
-    buildPrompt: (data) => `Draft a professional reply to this email:\n\n${(data as { body: string }).body}`,
-  });
-
-const server = new AgentServer({
-  port:          3000,
-  host:          '0.0.0.0',
-  engineFactory: enterprise.engineFactory,
-  auth:          new JwtAuth({ secret: process.env.JWT_SECRET!, allowQueryToken: false }).hook(),
-  strict:        true,
-  store:         new RedisSessionMetaStore(new Redis(process.env.REDIS_URL!)),
-  cors:          ['https://app.company.com'],
-  logger:        { level: 'info' },
-  rateLimiting:  { max: 60, timeWindow: '1 minute' },
-  metrics:       { enabled: true },
-  maxIdleMs:     30 * 60 * 1000,
-  triggers,
-});
-
-await server.start();
-process.on('SIGTERM', () => server.stop());
-```
+# @backendkit-labs/agent-web
+
+HTTP/WebSocket server layer for [@backendkit-labs/agent-core](https://www.npmjs.com/package/@backendkit-labs/agent-core).
+
+Wraps `AgentEngine` in a Fastify 5 server with streaming WebSocket sessions, JWT authentication, Prometheus metrics, rate limiting, a real-time monitor bus, and a flexible trigger system (cron jobs, webhooks, BullMQ queues).
+
+[![npm](https://img.shields.io/npm/v/@backendkit-labs/agent-web)](https://www.npmjs.com/package/@backendkit-labs/agent-web)
+
+## Table of contents
+
+- [Installation](#installation)
+- [Quick start](#quick-start)
+- [AgentServer options](#agentserver-options)
+- [HTTP and WebSocket routes](#http-and-websocket-routes)
+- [WebSocket protocol](#websocket-protocol)
+- [EngineFactory pattern](#enginefactory-pattern)
+- [SessionManager](#sessionmanager)
+- [JWT authentication](#jwt-authentication)
+- [Prometheus metrics](#prometheus-metrics)
+- [Rate limiting](#rate-limiting)
+- [Session metadata store](#session-metadata-store)
+- [MonitorBus](#monitorbus)
+- [TriggerBus — cron, webhooks, queues](#triggerbus--cron-webhooks-queues)
+- [HeadlessTransport](#headlesstransport)
+- [Full production example](#full-production-example)
+
+---
+
+## Installation
+
+```bash
+npm install @backendkit-labs/agent-web @backendkit-labs/agent-core
+```
+
+Optional integrations (install only what you need):
+
+```bash
+npm install jsonwebtoken          # JWT auth
+npm install prom-client           # Prometheus metrics
+npm install @fastify/rate-limit   # rate limiting
+npm install node-cron             # cron triggers
+npm install bullmq                # queue triggers
+npm install ioredis               # Redis session store
+```
+
+---
+
+## Quick start
+
+```typescript
+import { AgentServer } from '@backendkit-labs/agent-web';
+import type { EngineFactory } from '@backendkit-labs/agent-web';
+import { AgentEngine, AgentRegistry, ToolRegistry, ProviderRegistry } from '@backendkit-labs/agent-core';
+
+const engineFactory: EngineFactory = (sessionId, transport) =>
+  new AgentEngine({
+    transport,
+    sessionId,
+    // ...your engine config
+  });
+
+const server = new AgentServer({
+  port:          3000,
+  engineFactory,
+  cors:          '*',
+});
+
+await server.start();
+// WS  /chat      — streaming chat sessions
+// WS  /monitor   — real-time read-only observer
+// GET /health    — { status, sessions, monitors }
+// GET /sessions  — active session list
+// DELETE /sessions/:id — close a session
+```
+
+---
+
+## AgentServer options
+
+```typescript
+interface AgentServerOptions {
+  // Required
+  engineFactory: EngineFactory;       // creates a new AgentEngine per session
+
+  // Network
+  port?:   number;                    // default: 3000
+  host?:   string;                    // default: '127.0.0.1' (use '0.0.0.0', with auth, for remote callers)
+  cors?:   string | string[];         // default: '*' (none with strict)
+  strict?: boolean;                   // fail-closed defaults, see "Strict mode"
+  logger?: boolean | object;          // Fastify logger, default: false
+
+  // Auth
+  auth?: AuthHook;                    // called on every request before handlers
+
+  // Rate limiting (requires @fastify/rate-limit)
+  rateLimiting?: {
+    max?:        number;              // max requests per IP per window, default: 100
+    timeWindow?: string;             // e.g. '1 minute', '30 seconds'
+  };
+
+  // Prometheus metrics (requires prom-client)
+  metrics?: {
+    enabled:          boolean;        // must be true to activate
+    path?:            string;         // default: '/metrics'
+    prefix?:          string;         // metric name prefix
+    collectNodeMetrics?: boolean;     // include Node.js process metrics
+  };
+
+  // Session management
+  maxIdleMs?:   number;               // idle timeout before session is closed
+  maxSessions?: number;               // max concurrent sessions
+  store?:       SessionMetaStore;     // durable metadata store (default: in-memory)
+
+  // Triggers
+  triggers?: TriggerBus;             // wires webhook routes and starts cron jobs
+}
+```
+
+---
+
+## HTTP and WebSocket routes
+
+| Method | Path | Description |
+|--------|------|-------------|
+| `GET` | `/health` | Server health: `{ status, sessions, monitors, uptime }` |
+| `GET` | `/sessions` | Array of active session metadata |
+| `DELETE` | `/sessions/:id` | Force-close a session |
+| `GET` | `/metrics` | Prometheus metrics (if enabled) |
+| `WS` | `/chat` | Streaming chat — one session per connection |
+| `WS` | `/monitor` | Read-only observer of all sessions |
+| `POST` | `/webhooks/:name` | Webhook trigger (registered via TriggerBus) |
+
+---
+
+## WebSocket protocol
+
+### `/chat` — chat sessions
+
+**Client → server:**
+
+```json
+{ "type": "input",  "content": "Your message here" }
+{ "type": "abort" }
+```
+
+**Server → client** (streamed `AgentEvent` frames):
+
+```json
+{ "type": "run_start",     "sessionId": "abc-123" }
+{ "type": "token",         "content": "Hello" }
+{ "type": "tool_call",     "name": "search", "args": { "query": "..." } }
+{ "type": "tool_result",   "name": "search", "result": "..." }
+{ "type": "agent_switch",  "from": "general", "to": "security-reviewer" }
+{ "type": "done",          "final": "Full assistant response" }
+{ "type": "error",         "message": "Something went wrong" }
+```
+
+### `/monitor` — real-time observer
+
+Connect to observe all sessions without sending any messages:
+
+```json
+{ "type": "monitor_hello",         "sessionIds": ["abc-123", "def-456"] }
+{ "type": "monitor_session_open",  "sessionId": "ghi-789" }
+{ "type": "monitor_session_close", "sessionId": "ghi-789" }
+{ "sessionId": "abc-123", "type": "run_input", "content": "..." }
+{ "sessionId": "abc-123", "type": "token",     "content": "..." }
+{ "sessionId": "abc-123", "type": "done",      "final": "..." }
+```
+
+---
+
+## EngineFactory pattern
+
+`EngineFactory` is a function `(sessionId: string, transport: Transport) => AgentEngine`. Each incoming WebSocket connection gets its own isolated engine instance.
+
+### Minimal factory
+
+```typescript
+import type { EngineFactory } from '@backendkit-labs/agent-web';
+import { AgentEngine } from '@backendkit-labs/agent-core';
+
+const engineFactory: EngineFactory = (sessionId, transport) =>
+  new AgentEngine({
+    transport,
+    sessionId,
+    model:           { provider: 'openai', id: 'gpt-4o' },
+    agents,
+    tools,
+    providers,
+    defaultProvider: 'openai',
+    defaultAgentId:  'general',
+    maxIterations:   50,
+    iterationMode:   'auto',
+  });
+```
+
+### Factory with per-session memory
+
+```typescript
+import { createMemorySystem } from '@backendkit-labs/agent-core';
+import * as path from 'path';
+import * as os from 'os';
+
+const engineFactory: EngineFactory = (sessionId, transport) => {
+  // Isolated memory per session
+  const sessDir = path.join(os.homedir(), '.bk-agent', 'sessions', sessionId);
+  const memory = createMemorySystem({
+    episodic:  { provider: 'json', path: path.join(sessDir, 'episodes.json') },
+    semantic:  { provider: 'json', path: path.join(sessDir, 'knowledge.json') },
+    procedural:{ provider: 'json', path: path.join(sessDir, 'patterns.json') },
+  });
+
+  return new AgentEngine({
+    transport,
+    sessionId,
+    memorySystem: memory,
+    ...baseConfig,
+  });
+};
+```
+
+### Factory with per-user context
+
+Each WebSocket connection to `/chat` gets its own session with a random `sessionId`, and the factory builds a fresh engine for it. When `auth` is set, the authenticated user (the JWT `sub`) becomes the **owner** of the session: other users cannot see it in `/sessions` or `/monitor`. To load per-user settings in the engine, keep your own map from session to user, or read them inside your tools.
+
+```typescript
+const engineFactory: EngineFactory = (sessionId, transport) =>
+  new AgentEngine({
+    ...baseConfig,
+    transport,
+    sessionId,                 // also the session's id in metrics, monitor and audit
+  });
+```
+
+---
+
+## SessionManager
+
+`SessionManager` creates, tracks and cleans up sessions. `AgentServer` creates one internally; you can also use it on your own WebSocket server.
+
+```typescript
+import { SessionManager } from '@backendkit-labs/agent-web';
+
+const sessions = new SessionManager({
+  engineFactory,
+  maxIdleMs:   30 * 60 * 1000,  // close after 30 min of inactivity (default)
+  maxSessions: 100,             // reject new sessions beyond this (default)
+  monitorBus:  myMonitorBus,
+  metrics:     myMetricsCollector,
+  store:       myMetaStore,
+});
+
+const session = sessions.create(socket, userId);   // one per WebSocket connection; userId is the owner
+sessions.get(session.id);
+sessions.list();                                   // [{ id, createdAt, lastActivityAt, userId }]
+sessions.destroy(session.id);
+sessions.close();                                  // all sessions, on shutdown
+```
+
+---
+
+## JWT authentication
+
+`JwtAuth` validates Bearer tokens on HTTP routes and on the WebSocket handshake. Browsers cannot set headers on a WebSocket, so the token can also travel as the subprotocol `bearer, <jwt>`. A `?token=` query parameter still works unless `allowQueryToken: false`.
+
+### Strict mode
+
+For a server that other people reach, turn on both switches:
+
+```ts
+const auth = new JwtAuth({ secret: process.env.JWT_SECRET!, allowQueryToken: false }).hook();
+new AgentServer({ engineFactory, host: '0.0.0.0', auth, strict: true, cors: ['https://app.example.com'] });
+```
+
+- `allowQueryToken: false` ignores `?token=` (a token in a URL lands in logs, proxies and history). A browser sends the
+  token as the WebSocket subprotocol: `new WebSocket(url, ['bearer', jwt])`; `useAgentSession({ url, token })` in
+  `agent-web-ui` does it for you.
+- `strict: true` allows no CORS origin unless `cors` lists it, adds `X-Content-Type-Options`, `X-Frame-Options: DENY` and
+  `Referrer-Policy: no-referrer`, and refuses (throws) to listen on a non-loopback `host` without `auth`.
+
+### Setup
+
+```typescript
+import { AgentServer, JwtAuth } from '@backendkit-labs/agent-web';
+
+const jwt = new JwtAuth({
+  secret:     process.env.JWT_SECRET!,   // or publicKey for RS256 / ES256
+  algorithms: ['HS256'],                 // accepted algorithms
+  audience:   'agents',                  // optional checks
+  issuer:     'https://login.company.com',
+});
+
+const server = new AgentServer({ port: 3000, engineFactory, auth: jwt.hook() });
+```
+
+The token must carry `sub` (the user id, owner of the sessions it opens); `email` and `roles` are read when present. `adminRole` in `AgentServerOptions` names the role that may see every session.
+
+### Issuing tokens
+
+`JwtAuth` only verifies. Issue tokens in your login endpoint with any JWT library and the same secret or key pair:
+
+```typescript
+import jwt from 'jsonwebtoken';
+
+const token = jwt.sign({ sub: 'user-123', roles: ['admin'] }, process.env.JWT_SECRET!, { expiresIn: '24h' });
+// The client sends it as Authorization: Bearer <token>, or as the WebSocket subprotocol `bearer, <token>`.
+```
+
+### Custom auth hook
+
+If you use a different auth system (API keys, OAuth tokens), implement `AuthHook` directly:
+
+```typescript
+import type { AuthHook } from '@backendkit-labs/agent-web';
+
+const apiKeyAuth: AuthHook = async (req, reply) => {
+  const key = req.headers['x-api-key'];
+  if (!key || !validApiKeys.has(key as string)) {
+    reply.code(401).send({ error: 'Unauthorized' });
+  }
+};
+
+const server = new AgentServer({ auth: apiKeyAuth, ...opts });
+```
+
+---
+
+## Prometheus metrics
+
+Requires `prom-client` to be installed. Exposes a `/metrics` endpoint with standard Prometheus counters and histograms.
+
+```typescript
+const server = new AgentServer({
+  ...opts,
+  metrics: {
+    enabled:            true,
+    path:               '/metrics',         // default
+    prefix:             'bk_agent_',        // prepended to all metric names
+    collectNodeMetrics: true,               // include Node.js process metrics
+  },
+});
+```
+
+### Exposed metrics
+
+| Metric | Type | Description |
+|--------|------|-------------|
+| `bk_agent_sessions_active` | Gauge | Currently open sessions |
+| `bk_agent_runs_total{status}` | Counter | Total runs by status (`success`, `error`) |
+| `bk_agent_run_duration_seconds` | Histogram | Run latency distribution |
+| Node.js metrics | Various | Memory, CPU, GC (when `collectNodeMetrics: true`) |
+
+### Using MetricsCollector directly
+
+```typescript
+import { MetricsCollector } from '@backendkit-labs/agent-web';
+
+const collector = new MetricsCollector({ prefix: 'my_agent_' });
+
+// Record a run
+const endRun = collector.runStarted();
+// ... after it completes:
+endRun('success');   // or endRun('error')
+
+collector.sessionOpened();      // and sessionClosed(), messageReceived()
+
+// Prometheus text format (serve it with collector.contentType)
+const text = await collector.collect();
+// # HELP my_agent_runs_total ...
+// my_agent_runs_total{status="success"} 42
+```
+
+---
+
+## Rate limiting
+
+Requires `@fastify/rate-limit`. Applied globally to all routes:
+
+```typescript
+const server = new AgentServer({
+  ...opts,
+  rateLimiting: {
+    max:        60,          // max 60 requests per IP
+    timeWindow: '1 minute',  // sliding window
+  },
+});
+```
+
+When the limit is exceeded, the server responds with:
+```json
+{ "statusCode": 429, "error": "Too Many Requests", "message": "Rate limit exceeded" }
+```
+
+---
+
+## Session metadata store
+
+By default, session metadata lives in memory and is lost on restart. Use `RedisSessionMetaStore` for durable metadata across restarts and horizontal scaling.
+
+### In-memory (default)
+
+```typescript
+import { InMemorySessionMetaStore } from '@backendkit-labs/agent-web';
+
+const server = new AgentServer({
+  ...opts,
+  store: new InMemorySessionMetaStore(),
+});
+```
+
+### Redis store (requires ioredis or node-redis)
+
+`RedisSessionMetaStore` is duck-typed — it accepts any Redis client that implements `.get()`, `.set()`, and `.del()`:
+
+```typescript
+import { RedisSessionMetaStore } from '@backendkit-labs/agent-web';
+import Redis from 'ioredis';
+
+const redis = new Redis(process.env.REDIS_URL!);
+
+const server = new AgentServer({
+  ...opts,
+  store: new RedisSessionMetaStore(redis, {
+    ttlSeconds: 86400,    // keep session metadata for 24h
+    prefix:     'bk:sess:', // key prefix
+  }),
+});
+```
+
+---
+
+## MonitorBus
+
+`MonitorBus` broadcasts all session events to connected `/monitor` WebSocket observers. Used by the Kanban board in `@backendkit-labs/agent-web-ui`.
+
+### Standalone usage
+
+```typescript
+import { MonitorBus } from '@backendkit-labs/agent-web';
+
+const monitor = new MonitorBus();
+
+// Subscribe to all events (e.g. for logging or alerting)
+const unsub = monitor.subscribe((frame) => {
+  if (frame.type === 'done') {
+    console.log(`[${frame.sessionId}] run completed`);
+  }
+  if (frame.type === 'error') {
+    alerting.notify(`Agent error in session ${frame.sessionId}: ${frame.message}`);
+  }
+});
+
+// Clean up
+unsub();
+```
+
+---
+
+## TriggerBus — cron, webhooks, queues
+
+`TriggerBus` runs agents without a WebSocket client: on a schedule, on an incoming webhook, or for each job of a BullMQ queue. Each run gets a fresh engine from the factory and a `HeadlessTransport`, and shows up in the monitor.
+
+Every trigger has the same three callbacks:
+
+- `buildPrompt(payload)` returns the prompt. `payload` is `undefined` for cron, the parsed body for a webhook, and `job.data` for a queue job.
+- `onComplete(result)` receives `{ sessionId, triggerName, firedAt, durationMs, output, events }`.
+- `onError(err, triggerName)` is called when the run throws.
+
+### Cron trigger (requires `node-cron`)
+
+```typescript
+import { TriggerBus } from '@backendkit-labs/agent-web';
+
+const triggers = new TriggerBus(engineFactory)
+  .addCron({
+    name:        'daily-report',
+    schedule:    '0 8 * * 1-5',     // weekdays at 8 AM
+    buildPrompt: () => 'Generate the daily engineering standup summary from recent commits.',
+    runOnStart:  false,             // true fires once when the bus starts
+    onComplete:  ({ output }) => slack.send('#daily', output),
+    onError:     (err, name) => logger.error(`[${name}]`, err),
+  })
+  .addCron({
+    name:        'health-check',
+    schedule:    '*/15 * * * *',
+    buildPrompt: () => 'Check all API endpoints and report any that are down.',
+    onComplete:  ({ output }) => { if (output.includes('DOWN')) pagerduty.alert(output); },
+  });
+
+const server = new AgentServer({ ...opts, triggers });
+await server.start();   // cron jobs start with the server
+```
+
+### Webhook trigger
+
+```typescript
+const triggers = new TriggerBus(engineFactory)
+  .addWebhook({
+    name:        'github-pr',
+    path:        '/webhooks/github',          // POST /webhooks/github (method: 'GET' | 'POST' | 'PUT')
+    secret:      process.env.GITHUB_SECRET,   // HMAC X-Hub-Signature-256 over the raw body
+    buildPrompt: (body) => {
+      const pr = (body as { pull_request: { number: number; title: string; diff_url: string } }).pull_request;
+      return `Review PR #${pr.number}: "${pr.title}"\n${pr.diff_url}`;
+    },
+    onComplete:  ({ output }) => github.comment(output),
+  });
+```
+
+The webhook answers immediately and the run continues in the background; use `onComplete` to act on the result. If the payload is needed there, capture it in `buildPrompt`.
+
+```bash
+curl -X POST http://localhost:3000/webhooks/github \
+  -H "Content-Type: application/json" \
+  -H "X-Hub-Signature-256: sha256=<hmac of the exact body>" \
+  -d '{ "pull_request": { "number": 42, "title": "Add auth module", "diff_url": "..." } }'
+```
+
+### Queue trigger (BullMQ — requires `bullmq`)
+
+```typescript
+const triggers = new TriggerBus(engineFactory)
+  .addQueue({
+    name:        'document-processor',
+    queueName:   'documents',
+    connection:  process.env.REDIS_URL!,   // URL, or { host, port, password, username, db, tls }
+    concurrency: 3,                        // default 1
+    buildPrompt: (data) => `Summarize and extract key points:\n\n${(data as { content: string }).content}`,
+    onComplete:  ({ output }) => saveSummary(output),
+  });
+```
+
+A failed run is rethrown to BullMQ, so the job is marked failed and its retry policy applies. Jobs are published with BullMQ as usual (`new Queue('documents', { connection }).add('process', { id, content }, { attempts: 3 })`).
+
+---
+
+## HeadlessTransport
+
+A `Transport` with no socket: it buffers every `AgentEvent` in memory. `TriggerBus` uses it for autonomous runs; use it to run an engine from a script.
+
+```typescript
+import { HeadlessTransport } from '@backendkit-labs/agent-web';
+
+const transport = new HeadlessTransport();
+const engine = new AgentEngine({ ...opts, transport });
+await engine.run('Generate a daily report');
+
+transport.getOutput();   // concatenated text of the 'token' events
+transport.getEvents();   // every event, in order
+```
+
+---
+
+## Full production example
+
+JWT auth, Redis sessions, Prometheus metrics, a cron job, a webhook and a queue:
+
+```typescript
+import { AgentServer, JwtAuth, TriggerBus, RedisSessionMetaStore } from '@backendkit-labs/agent-web';
+import { createEnterpriseSetup } from '@backendkit-labs/agent-enterprise';
+import { ProviderRegistry } from '@backendkit-labs/agent-core';
+import { DeepSeekProvider } from '@backendkit-labs/agent-providers';
+import Redis from 'ioredis';
+
+const providers = new ProviderRegistry();
+providers.register('deepseek', new DeepSeekProvider({ apiKey: process.env.DEEPSEEK_API_KEY!, model: 'deepseek-chat' }));
+
+const enterprise = createEnterpriseSetup({ vaultPath: process.env.VAULT_PATH!, providers, defaultProvider: 'deepseek' });
+await enterprise.indexAll();
+
+const triggers = new TriggerBus(enterprise.engineFactory)
+  .addCron({
+    name:        'weekly-digest',
+    schedule:    '0 9 * * 1',
+    buildPrompt: () => 'Generate a weekly digest of all knowledge base changes.',
+    onComplete:  ({ output }) => slack.send('#general', output),
+  })
+  .addWebhook({
+    name:        'support-ticket',
+    path:        '/webhooks/support',
+    secret:      process.env.WEBHOOK_SECRET,
+    buildPrompt: (body) => {
+      const t = body as { id: string; subject: string; description: string };
+      return `Handle support ticket #${t.id}: ${t.subject}\n${t.description}`;
+    },
+  })
+  .addQueue({
+    name:        'email-processor',
+    queueName:   'emails',
+    connection:  process.env.REDIS_URL!,
+    concurrency: 5,
+    buildPrompt: (data) => `Draft a professional reply to this email:\n\n${(data as { body: string }).body}`,
+  });
+
+const server = new AgentServer({
+  port:          3000,
+  host:          '0.0.0.0',
+  engineFactory: enterprise.engineFactory,
+  auth:          new JwtAuth({ secret: process.env.JWT_SECRET!, allowQueryToken: false }).hook(),
+  strict:        true,
+  store:         new RedisSessionMetaStore(new Redis(process.env.REDIS_URL!)),
+  cors:          ['https://app.company.com'],
+  logger:        { level: 'info' },
+  rateLimiting:  { max: 60, timeWindow: '1 minute' },
+  metrics:       { enabled: true },
+  maxIdleMs:     30 * 60 * 1000,
+  triggers,
+});
+
+await server.start();
+process.on('SIGTERM', () => server.stop());
+```

package/package.json

@@ -1,51 +1,51 @@
-{
-  "name": "@backendkit-labs/agent-web",
-  "version": "0.6.5",
-  "description": "HTTP/WebSocket server layer for agent-core — browser-ready streaming agent sessions",
-  "main": "dist/index.js",
-  "types": "dist/index.d.ts",
-  "files": [
-    "dist/",
-    "client/",
-    "README.md"
-  ],
-  "scripts": {
-    "build": "tsc",
-    "dev": "tsc --watch",
-    "typecheck": "tsc --noEmit",
-    "test": "jest",
-    "prepublishOnly": "npm run build"
-  },
-  "peerDependencies": {
-    "@backendkit-labs/agent-core": ">=0.24.0 <0.40.0"
-  },
-  "dependencies": {
-    "@fastify/websocket": "^11.0.0",
-    "fastify": "^5.0.0",
-    "jsonwebtoken": "^9.0.0",
-    "prom-client": "^15.0.0"
-  },
-  "optionalDependencies": {
-    "@fastify/rate-limit": "^9.0.0",
-    "bullmq": "^5.0.0",
-    "node-cron": "^4.2.1"
-  },
-  "devDependencies": {
-    "@backendkit-labs/agent-core": "^0.38.0",
-    "@types/jest": "^29.5.14",
-    "@types/jsonwebtoken": "^9.0.0",
-    "@types/node": "^20.0.0",
-    "@types/ws": "^8.0.0",
-    "bullmq": "^5.0.0",
-    "jest": "^29.7.0",
-    "ts-jest": "^29.4.11",
-    "typescript": "^5.4.0",
-    "ws": "^8.21.0"
-  },
-  "engines": {
-    "node": ">=20.0.0"
-  },
-  "publishConfig": {
-    "access": "public"
-  }
-}
+{
+  "name": "@backendkit-labs/agent-web",
+  "version": "0.6.6",
+  "description": "HTTP/WebSocket server layer for agent-core — browser-ready streaming agent sessions",
+  "main": "dist/index.js",
+  "types": "dist/index.d.ts",
+  "files": [
+    "dist/",
+    "client/",
+    "README.md"
+  ],
+  "scripts": {
+    "build": "tsc",
+    "dev": "tsc --watch",
+    "typecheck": "tsc --noEmit",
+    "test": "jest",
+    "prepublishOnly": "npm run build"
+  },
+  "peerDependencies": {
+    "@backendkit-labs/agent-core": ">=0.24.0 <0.41.0"
+  },
+  "dependencies": {
+    "@fastify/websocket": "^11.0.0",
+    "fastify": "^5.0.0",
+    "jsonwebtoken": "^9.0.0",
+    "prom-client": "^15.0.0"
+  },
+  "optionalDependencies": {
+    "@fastify/rate-limit": "^9.0.0",
+    "bullmq": "^5.0.0",
+    "node-cron": "^4.2.1"
+  },
+  "devDependencies": {
+    "@backendkit-labs/agent-core": "^0.39.0",
+    "@types/jest": "^29.5.14",
+    "@types/jsonwebtoken": "^9.0.0",
+    "@types/node": "^20.0.0",
+    "@types/ws": "^8.0.0",
+    "bullmq": "^5.0.0",
+    "jest": "^29.7.0",
+    "ts-jest": "^29.4.11",
+    "typescript": "^5.4.0",
+    "ws": "^8.21.0"
+  },
+  "engines": {
+    "node": ">=20.0.0"
+  },
+  "publishConfig": {
+    "access": "public"
+  }
+}