@tailored-ai/server

HTTP API server for Tailored AI. Hono-based routes for sessions, agents, workflows, tasks, memory, and resources, plus SSE for streaming responses and webhook endpoints for cron / external triggers. Optionally serves a static UI build alongside the API.

This is the HTTP surface that the tai CLI and the @tailored-ai/ui frontend both talk to. You can embed it in your own Node process to expose your agent over HTTP.

npm install @tailored-ai/server @tailored-ai/core

Quick start

import Database from "better-sqlite3";
import {
  AgentRuntime,
  createProvider,
  loadConfig,
} from "@tailored-ai/core";
import { startServer } from "@tailored-ai/server";

const config = await loadConfig("./config.yaml");
const db = new Database("./agent.db");
const provider = createProvider(config);
const runtime = new AgentRuntime({ config, db, provider });

await startServer({ runtime, port: 3000 });
// GET  /api/agents
// POST /api/sessions/:id/messages   (SSE stream)
// POST /api/workflows/:name/run
// GET  /                            (UI, if uiDist provided)

To serve the bundled UI, pass uiDist pointing at a built SPA directory (the CLI does this automatically with the UI bundle it ships).

What's exposed

• /api/agents, /api/sessions, /api/messages — chat surface
• /api/workflows/... — workflow CRUD + run endpoints
• /api/tasks, /api/projects — task and project management
• /api/memory/... — memory recall, search, promotion
• /webhooks/... — external triggers (Stripe, GitHub, custom)
• /sse/... — streaming response channel

See src/index.ts for the full route table.

Docs

docs/architecture.md covers the integration model and how the server hangs off AgentRuntime.

License

MIT.