Open-Source Screen Recording for AI + Humans

Screen recordings your AI agents can understand

Record your screen, share with your team, and let AI agents extract structured context — timelines, actions, and transcripts — automatically.

app.example.com/dashboard

REC

Built for developers working with

ClaudeOpenAILangChainCursorWindsurfCrewAI

How Iris works

From screen recording to AI-consumable context in three steps.

Capture any screen

Use the browser extension, CLI, or upload any video. Iris captures everything — your screen, mic audio, and browser actions — in one click.

Recording...

Tab: Dashboard - Settings

02:34

Browser Extension

CLI Upload

Direct Upload

Everything you need

Screen recording, AI context extraction, and sharing — in one open-source platform.

AI Vision Pipeline

Every frame analyzed by Claude to extract what's happening on screen — OCR text, UI state, active applications, and more.

Structured Context

Timelines, actions, transcripts, URLs, code snippets, and tags — all machine-readable and queryable.

MCP Server

Connect directly to Claude Desktop, Cursor, or any MCP-compatible AI agent. One config line.

Public Sharing

Share recordings with a link — like Loom, but open-source. Video player with context tabs included.

REST API + SDK

Full REST API, TypeScript SDK, and CLI. Integrate with any language or framework in minutes.

Self-Hostable

Deploy on your own infrastructure with Docker. Your data stays on your servers. 100% open-source.

Built for real workflows

Whether you're recording for your team or your AI agents.

QA Bug Reports

Record the bug as you encounter it
Share the link with your team — no meeting needed
AI extracts exact repro steps, URLs, and error messages
Agents can query the context to auto-file detailed tickets

Developer Workflows

Record your workflow once, share the knowledge forever
AI agents learn your patterns from structured context
Search across all recordings: "how did I deploy last time?"
Build automations from recorded action sequences

Agent-Assisted Testing

Agents record test runs via CLI or SDK
Each recording produces structured pass/fail context
Timeline events pinpoint exactly where failures happen
MCP server lets agents review test results natively

Integrate in minutes

Connect your AI agents via MCP, CLI, or SDK.

// claude_desktop_config.json
{
  "mcpServers": {
    "iris": {
      "command": "npx",
      "args": ["@iris/mcp-server"],
      "env": {
        "IRIS_API_KEY": "ir_live_xxxxx"
      }
    }
  }
}

// Then just tell Claude:
// "Check the most recent recording"