Synento documentation

Synento is an agent-native video API. A coding agent can set up a project, write the integration, and prove media arrived by reading the session manifest. Live video, replay, and MP4 export all read the same durable streams.

Quick links

Getting started: set up your first project and stream in minutes
Core concepts: projects, sessions, and the stream-native recording model
API reference: every endpoint, with request and response examples

SDK packages

Package	Use
`@synento/client`	Browser: publish and receive live audio, video, and screen share
`@synento/react`	React provider, hooks, and components over `@synento/client`
`@synento/node`	Server: create sessions, mint connection tokens, manage projects
`@synento/player`	Embeddable replay engine with layout composition
`@synento/cli`	Command-line project and session management

Guides

Webhooks: subscribe to session lifecycle events
Recordings & Replay: how stream-based recording and replay work
Edited clips with JSON2Video: turn a recording into a short vertical clip with subtitles
Access Control: public sessions, expiring links, and visibility
Session Management: session lifecycle and controls
CLI Guide: use the Synento CLI for project and session management
SDK Guide: integrate using the JavaScript/TypeScript SDK
Browser Support: which browsers work, what the SDK reports, and what has been verified

Examples

Copy-paste examples ready to run:

Web Quickstart: embed a live video stream in any web page
SDK Basic Usage: create sessions and manage them with the SDK
SDK Advanced Patterns: replay, exports, and webhooks with the SDK
CLI Examples: common CLI workflows

Key features

Feature	Description
Live streaming	Multi-participant audio and video over WebSocket, encoded in the browser with WebCodecs; frame rate backs off on a congested connection
Stream-native recording	Every session is recorded as it happens: separated tracks (cam/mic/screen) plus a typed event timeline, layout-agnostic
Composable replay	Configurable layouts (grid, active-speaker, presenter, screen-share) at playback time
Director track	Deterministic active-speaker switching, computed when a session ends and stored with the recording
Replay	Replay with speed control (0.25x-4x), seek, layout switching, and windowed reading
MP4 export	Async export with layout (`grid`, `presenter`) and orientation (`landscape`, `vertical`) parameters
Public sessions and links	Share sessions publicly or via expiring token-based replay links
Webhooks	Event-driven callbacks for session lifecycle events
SDK and CLI	TypeScript SDK and command-line tool for programmatic access

API overview

Everything goes through REST APIs. Authenticate with an API key in the Authorization: Bearer sk_live_... header. Live media travels over a WebSocket, using a connection token minted by the REST API.

All JSON responses follow a standard envelope format:

{ "data": { ... }, "error": null }           // Success
{ "data": null, "error": { "code": "...", "message": "...", "status": 400 } }  // Error

Key endpoint categories:

POST /v1/projects: create projects
POST /v1/sessions: create live sessions
POST /v1/sessions/:id/connection: get connection tokens for WebSocket streaming
GET /v1/sessions/:id/manifest: tracks, time bounds, and capabilities
GET /v1/sessions/:id/playback: windowed replay of recorded sessions
GET /v1/sessions/:id/export: enqueue MP4 export (layout, orientation)
GET /v1/exports/:jobId: poll export status and download URL

See the full API reference for every endpoint, or the machine-readable spec at synento.com/openapi.json.