@mastra/mcp-docs-server 1.1.5 → 1.1.6-alpha.0

package/.docs/docs/server/server-adapters.md

@@ -0,0 +1,547 @@
+# Server Adapters
+
+Server adapters let you run Mastra with your own HTTP server instead of the Hono server generated by `mastra build`. They provide more control over the server setup, including custom middleware ordering, authentication, logging, and deployment configuration. You can still integrate Mastra into any Node.js application without changing how agents or workflows execute.
+
+## When to use server adapters
+
+- You want Mastra’s endpoints added automatically to an existing application
+- You need direct access to the server instance for custom configuration
+- Your team prefers using another server framework instead of the Hono server created by `mastra build`.
+
+> **Tip:** For deployments without custom server requirements, use `mastra build` instead. It configures server setup, registers middleware, and applies deployment settings based on your project configuration. See [Server Configuration](https://mastra.ai/docs/server/mastra-server).
+>
+> If you want to use [Studio](https://mastra.ai/docs/getting-started/studio) with your server adapter, use [`mastra studio`](https://mastra.ai/reference/cli/mastra) to only launch the Studio UI.
+
+## Available adapters
+
+Mastra currently provides these official server adapters:
+
+- [@mastra/express](https://mastra.ai/reference/server/express-adapter)
+- [@mastra/hono](https://mastra.ai/reference/server/hono-adapter)
+- [@mastra/fastify](https://mastra.ai/reference/server/fastify-adapter)
+- [@mastra/koa](https://mastra.ai/reference/server/koa-adapter)
+
+You can build your own adapter, read [custom adapters](https://mastra.ai/docs/server/custom-adapters) for details.
+
+## Installation
+
+Install the adapter for the framework of your choice.
+
+**Express**:
+
+**npm**:
+
+```bash
+npm install @mastra/express@latest
+```
+
+**pnpm**:
+
+```bash
+pnpm add @mastra/express@latest
+```
+
+**Yarn**:
+
+```bash
+yarn add @mastra/express@latest
+```
+
+**Bun**:
+
+```bash
+bun add @mastra/express@latest
+```
+
+**Hono**:
+
+```bash
+npm install @mastra/express@latest
+```
+
+**Fastify**:
+
+```bash
+pnpm add @mastra/express@latest
+```
+
+**Koa**:
+
+```bash
+yarn add @mastra/express@latest
+```
+
+**Tab 5**:
+
+```bash
+bun add @mastra/express@latest
+```
+
+**Tab 6**:
+
+**npm**:
+
+```bash
+npm install @mastra/hono@latest
+```
+
+**pnpm**:
+
+```bash
+pnpm add @mastra/hono@latest
+```
+
+**Yarn**:
+
+```bash
+yarn add @mastra/hono@latest
+```
+
+**Bun**:
+
+```bash
+bun add @mastra/hono@latest
+```
+
+**Tab 7**:
+
+```bash
+npm install @mastra/hono@latest
+```
+
+**Tab 8**:
+
+```bash
+pnpm add @mastra/hono@latest
+```
+
+**Tab 9**:
+
+```bash
+yarn add @mastra/hono@latest
+```
+
+**Tab 10**:
+
+```bash
+bun add @mastra/hono@latest
+```
+
+**Tab 11**:
+
+**npm**:
+
+```bash
+npm install @mastra/fastify@latest
+```
+
+**pnpm**:
+
+```bash
+pnpm add @mastra/fastify@latest
+```
+
+**Yarn**:
+
+```bash
+yarn add @mastra/fastify@latest
+```
+
+**Bun**:
+
+```bash
+bun add @mastra/fastify@latest
+```
+
+**Tab 12**:
+
+```bash
+npm install @mastra/fastify@latest
+```
+
+**Tab 13**:
+
+```bash
+pnpm add @mastra/fastify@latest
+```
+
+**Tab 14**:
+
+```bash
+yarn add @mastra/fastify@latest
+```
+
+**Tab 15**:
+
+```bash
+bun add @mastra/fastify@latest
+```
+
+**Tab 16**:
+
+**npm**:
+
+```bash
+npm install @mastra/koa@latest
+```
+
+**pnpm**:
+
+```bash
+pnpm add @mastra/koa@latest
+```
+
+**Yarn**:
+
+```bash
+yarn add @mastra/koa@latest
+```
+
+**Bun**:
+
+```bash
+bun add @mastra/koa@latest
+```
+
+**Tab 17**:
+
+```bash
+npm install @mastra/koa@latest
+```
+
+**Tab 18**:
+
+```bash
+pnpm add @mastra/koa@latest
+```
+
+**Tab 19**:
+
+```bash
+yarn add @mastra/koa@latest
+```
+
+**Tab 20**:
+
+```bash
+bun add @mastra/koa@latest
+```
+
+## Configuration
+
+Initialize your app as usual, then create a `MastraServer` by passing in the `app` and your main `mastra` instance from `src/mastra/index.ts`. Calling `init()` automatically registers Mastra middleware and all available endpoints. You can continue adding your own routes as normal, either before or after `init()`, and they’ll run alongside Mastra’s endpoints.
+
+**Express**:
+
+```typescript
+import express from 'express'
+import { MastraServer } from '@mastra/express'
+import { mastra } from './mastra'
+
+const app = express()
+app.use(express.json())
+
+const server = new MastraServer({ app, mastra })
+
+await server.init()
+
+app.listen(4111, () => {
+  console.log('Server running on port 4111')
+})
+```
+
+> **Info:** See the [Express Adapter](https://mastra.ai/reference/server/express-adapter) documentation for full configuration options.
+
+**Hono**:
+
+```typescript
+import { Hono } from 'hono'
+import { serve } from '@hono/node-server'
+import { HonoBindings, HonoVariables, MastraServer } from '@mastra/hono'
+import { mastra } from './mastra'
+
+const app = new Hono<{ Bindings: HonoBindings; Variables: HonoVariables }>()
+
+const server = new MastraServer({ app, mastra })
+
+await server.init()
+
+serve({ fetch: app.fetch, port: 4111 }, () => {
+  console.log('Server running on port 4111')
+})
+```
+
+> **Info:** See the [Hono Adapter](https://mastra.ai/reference/server/hono-adapter) documentation for full configuration options.
+
+**Fastify**:
+
+```typescript
+import Fastify from 'fastify'
+import { MastraServer } from '@mastra/fastify'
+import { mastra } from './mastra'
+
+const app = Fastify()
+const server = new MastraServer({ app, mastra })
+
+await server.init()
+
+app.get('/health', async request => {
+  const mastraInstance = request.mastra
+  const agents = Object.keys(mastraInstance.listAgents())
+  return { status: 'ok', agents }
+})
+
+const port = 4111
+
+app.listen({ port }, () => {
+  console.log(`Server running on http://localhost:${port}`)
+  console.log(`Try: curl http://localhost:${port}/api/agents`)
+})
+```
+
+> **Info:** See the [Fastify Adapter](https://mastra.ai/reference/server/fastify-adapter) documentation for full configuration options.
+
+**Koa**:
+
+```typescript
+import Koa from 'koa'
+import bodyParser from 'koa-bodyparser'
+import { MastraServer } from '@mastra/koa'
+import { mastra } from './mastra'
+
+const app = new Koa()
+app.use(bodyParser()) // Required for body parsing
+
+const server = new MastraServer({ app, mastra })
+
+await server.init()
+
+app.use(async (ctx, next) => {
+  if (ctx.path === '/health' && ctx.method === 'GET') {
+    const mastraInstance = ctx.state.mastra
+    const agents = Object.keys(mastraInstance.listAgents())
+    ctx.body = { status: 'ok', agents }
+    return
+  }
+  await next()
+})
+
+const port = 4111
+
+app.listen(port, () => {
+  console.log(`Server running on http://localhost:${port}`)
+  console.log(`Try: curl http://localhost:${port}/api/agents`)
+})
+```
+
+> **Info:** See the [Koa Adapter](https://mastra.ai/reference/server/koa-adapter) documentation for full configuration options.
+
+## Initialization flow
+
+Calling `init()` runs three steps in order. Understanding this flow helps when you need to insert your own middleware at specific points.
+
+1. `registerContextMiddleware()`: Attaches the Mastra instance, request context, tools, and abort signal to every request. This makes Mastra available to all subsequent middleware and route handlers.
+2. `registerAuthMiddleware()`: Adds authentication and authorization middleware, but only if `server.auth` is configured in your Mastra instance. Skipped entirely if no auth is configured.
+3. `registerRoutes()`: Registers all Mastra API routes for agents, workflows, and other features. Also registers MCP routes if MCP servers are configured.
+
+### Manual initialization
+
+For custom middleware ordering, call each method separately instead of `init()`. This is useful when you need middleware that runs before Mastra's context is set up, or when you need to insert logic between the initialization steps.
+
+```typescript
+const server = new MastraServer({ app, mastra });
+
+// Your middleware first
+app.use(loggingMiddleware);
+
+server.registerContextMiddleware();
+
+// Middleware that needs Mastra context
+app.use(customMiddleware);
+
+server.registerAuthMiddleware();
+await server.registerRoutes();
+
+// Routes after Mastra
+app.get('/health', ...);
+```
+
+> **Tip:** Use manual initialization when you need middleware that runs before Mastra's context is available, or when you need to insert middleware between the context and auth steps.
+
+## Adding custom routes
+
+You can add your own routes to the app alongside Mastra's routes.
+
+- Routes added **before** `init()` won't have Mastra context available.
+- Routes added **after** `init()` have access to the Mastra context (the Mastra instance, request context, authenticated user, etc.).
+
+> **Info:** Visit "Adding custom routes" for [Express](https://mastra.ai/reference/server/express-adapter) and [Hono](https://mastra.ai/reference/server/hono-adapter) for more information.
+
+## Route prefixes
+
+By default, Mastra routes are registered at `/api/agents`, `/api/workflows`, etc. Use the `prefix` option to change this. This is useful for API versioning or when integrating with an existing app that has its own `/api` routes.
+
+```typescript
+const server = new MastraServer({
+  app,
+  mastra,
+  prefix: '/api/v2',
+})
+```
+
+With this prefix, Mastra routes become `/api/v2/agents`, `/api/v2/workflows`, etc. Custom routes you add directly to the app are not affected by this prefix.
+
+## OpenAPI spec
+
+Mastra can generate an OpenAPI specification for all registered routes. This is useful for documentation, client generation, or integration with API tools. Enable it by setting the `openapiPath` option:
+
+```typescript
+const server = new MastraServer({
+  app,
+  mastra,
+  openapiPath: '/openapi.json',
+})
+```
+
+The spec is generated from the Zod schemas defined on each route and served at the specified path. It includes all Mastra routes as well as any custom routes created with `createRoute()`.
+
+## Stream data redaction
+
+When streaming agent responses over HTTP, the HTTP streaming layer redacts sensitive information from stream chunks before sending them to clients. This prevents accidental exposure of:
+
+- System prompts and agent instructions
+- Tool definitions and their parameters
+- API keys and other credentials in request bodies
+- Internal configuration data
+
+This redaction happens at the HTTP boundary, so internal callbacks like `onStepFinish` still have access to the full request data for debugging and observability purposes.
+
+By default, redaction is enabled. Configure this behavior via `streamOptions`. Set `redact: false` only for internal services or debugging scenarios where you need access to the full request data in stream responses.
+
+```typescript
+const server = new MastraServer({
+  app,
+  mastra,
+  streamOptions: {
+    redact: true, // Default
+  },
+})
+```
+
+> **Info:** See [MastraServer](https://mastra.ai/reference/server/mastra-server) for full configuration options.
+
+## Per-route auth overrides
+
+When authentication is configured on your Mastra instance, all routes require authentication by default. Sometimes you need exceptions: public health check endpoints, webhook receivers, or admin routes that need stricter controls.
+
+Use `customRouteAuthConfig` to override authentication behavior for specific routes. Keys follow the format `METHOD:PATH` where method is `GET`, `POST`, `PUT`, `DELETE`, or `ALL`. Paths support wildcards (`*`) for matching multiple routes. Setting a value to `false` makes the route public, while `true` requires authentication.
+
+```typescript
+const server = new MastraServer({
+  app,
+  mastra,
+  customRouteAuthConfig: new Map([
+    // Public health check
+    ['GET:/api/health', false],
+    // Public API spec
+    ['GET:/api/openapi.json', false],
+    // Public webhook endpoints
+    ['POST:/api/webhooks/*', false],
+    // Require auth even if globally disabled
+    ['POST:/api/admin/reset', true],
+    // Protect all methods on internal routes
+    ['ALL:/api/internal/*', true],
+  ]),
+})
+```
+
+> **Info:** See [MastraServer](https://mastra.ai/reference/server/mastra-server) for full configuration options.
+
+## Accessing the app
+
+After creating the adapter, you may still need access to the underlying framework app. This is useful when passing it to a platform’s `serve` function or when adding routes from another module.
+
+```typescript
+// Via the MastraServer instance
+const app = server.getApp()
+
+// Via the Mastra instance (available after adapter construction)
+const app = mastra.getServerApp()
+```
+
+Both methods return the same app instance. Use whichever is more convenient based on what's in scope.
+
+## Server config vs adapter options
+
+When using server adapters, configuration comes from two places: the Mastra `server` config (passed to the `Mastra` constructor) and the adapter constructor options. Understanding which options come from where helps avoid confusion when settings don't seem to take effect.
+
+### Used by adapters
+
+The adapter reads these settings from `mastra.getServer()`:
+
+| Option          | Description                                                                                                                                     |
+| --------------- | ----------------------------------------------------------------------------------------------------------------------------------------------- |
+| `auth`          | Authentication config, used by `registerAuthMiddleware()`.                                                                                      |
+| `bodySizeLimit` | Default body size limit in bytes. Can be overridden per-adapter via `bodyLimitOptions`.                                                         |
+| `onError`       | Custom error handler called when an unhandled error occurs in a route handler. See [server.onError](https://mastra.ai/reference/configuration). |
+
+### Adapter constructor only
+
+These options are passed directly to the adapter constructor and are not read from the Mastra config:
+
+| Option                  | Description                                                                 |
+| ----------------------- | --------------------------------------------------------------------------- |
+| `prefix`                | Route path prefix                                                           |
+| `openapiPath`           | OpenAPI spec endpoint                                                       |
+| `bodyLimitOptions`      | Body size limit with custom error handler                                   |
+| `streamOptions`         | Stream redaction settings                                                   |
+| `customRouteAuthConfig` | Per-route auth overrides                                                    |
+| `mcpOptions`            | MCP transport options (e.g., `serverless: true` for stateless environments) |
+
+### Not used by adapters
+
+These `server` config options are only used by `mastra build` and have no effect when using adapters directly:
+
+| Option         | Used by                                 |
+| -------------- | --------------------------------------- |
+| `port`, `host` | `mastra dev`, `mastra build`            |
+| `cors`         | `mastra build` adds CORS middleware     |
+| `timeout`      | `mastra build`                          |
+| `apiRoutes`    | `registerApiRoute()` for `mastra build` |
+| `middleware`   | Middleware config for `mastra build`    |
+
+When using adapters, configure these features directly with your framework. For example, add CORS middleware using Hono's or Express's built-in CORS packages, and set the port when calling your framework's listen function.
+
+## MCP support
+
+Server adapters register MCP (Model Context Protocol) routes during `registerRoutes()` when MCP servers are configured in your Mastra instance. MCP allows external tools and services to connect to your Mastra server and interact with your agents.
+
+The adapter registers routes for both HTTP and SSE (Server-Sent Events) transports, enabling different client connection patterns.
+
+### Serverless mode
+
+For serverless environments like Cloudflare Workers or Vercel Edge, enable stateless mode via `mcpOptions`:
+
+```typescript
+const server = new MastraServer({
+  app,
+  mastra,
+  mcpOptions: {
+    serverless: true,
+  },
+})
+```
+
+When `serverless: true`, MCP HTTP requests run without session management, making them compatible with stateless execution environments.
+
+See [MCP](https://mastra.ai/docs/mcp/overview) for configuration details and how to set up MCP servers.
+
+## Related
+
+- [Hono Adapter](https://mastra.ai/reference/server/hono-adapter) - Hono-specific setup
+- [Express Adapter](https://mastra.ai/reference/server/express-adapter) - Express-specific setup
+- [Custom Adapters](https://mastra.ai/docs/server/custom-adapters) - Building adapters for other frameworks
+- [Server Configuration](https://mastra.ai/docs/server/mastra-server) - Using `mastra build` instead
+- [Authentication](https://mastra.ai/docs/server/auth) - Configuring auth for your server
+- [MastraServer Reference](https://mastra.ai/reference/server/mastra-server) - Full API reference
+- [createRoute() Reference](https://mastra.ai/reference/server/create-route) - Creating type-safe custom routes