{
  "markdown": "# ThinkRun — Screen Recorder for AI Coding Agents\n\nStop screenshotting bugs and typing paragraphs to explain them. ThinkRun records any browser session — or lets your agent record what it built — and turns it into **structured context your AI coding agent can act on**: every click, the console, network requests, your voice narration, and screenshots, all time-synced. Not just a screenshot. Not just a video.\n\nWorks with **Claude Code**, **Cursor**, **Cline**, **Windsurf**, and any MCP client.\n\n## Install\n\n```bash\nnpx @thinkrun/mcp\n```\n\nOr install the CLI globally:\n\n```bash\nnpm install -g @thinkrun/cli\n```\n\nThe CLI automatically installs the native host binary, which lets AI tools control Chrome.\n\n## Add to your MCP config\n\n```json\n{\n  \"mcpServers\": {\n    \"thinkrun\": {\n      \"command\": \"npx\",\n      \"args\": [\"@thinkrun/mcp\"]\n    }\n  }\n}\n```\n\nConfig location:\n- **Claude Code**: `~/.claude/settings.json`\n- **Cursor**: `.cursor/mcp.json`\n- **Cline**: VS Code settings → Cline MCP Servers\n\nThen ask your agent: *\"navigate to localhost:3000 and screenshot the checkout page\"* — or record the bug yourself and hand the agent the result.\n\n## Why ThinkRun\n\nOther tools give your agent eyes. ThinkRun gives it **structured sight** — context it can parse in one pass instead of you re-explaining the page.\n\n| | ThinkRun | Playwright MCP |\n|---|---|---|\n| Records a session into structured, replayable context | ✅ | ❌ drive-only |\n| Turns a recording into an AI-ready report (clicks, console, network) | ✅ | ❌ |\n| Shareable link, also as LLM-ready Markdown / JSON | ✅ | ❌ |\n| Uses your real Chrome — your cookies & sessions | ✅ | ❌ headless only |\n| Runs alongside your browser, no profile conflict | ✅ | ❌ fights for Chrome profile |\n\n## Agent skills for Claude Code, Cursor, Codex & Gemini CLI\n\nFour portable [agent skills](https://agentskills.io) ship in this repo, mirrored for every major harness (`.claude/skills/`, `.cursor/skills/`, `.codex/skills/`, `.gemini/skills/`):\n\n- **web-browse** — browse the web programmatically: drive a real or cloud browser to navigate, interact, extract, and screenshot.\n- **ux-audit** — walk through a product UI as a real user, screenshot every step, and produce a structured UX report with a prioritized fix list.\n- **thinkbrowse-cli** — drive the browser from shell scripts via the `thinkrun` CLI.\n- **thinkbrowse-mcp** — drive the browser through the ThinkRun MCP server's tools.\n\n**Install without cloning** — the repo is served as a live MCP server that auto-updates on every commit:\n\n```json\n{ \"mcpServers\": { \"thinkrun-skills\": { \"url\": \"https://mcp.skillsovermcp.com/mcp/dundas/thinkrun\" } } }\n```\n\n**Trigger accuracy is measured, not claimed.** Every skill ships a 20-query trigger-eval fixture, and [`evals/BENCHMARK.md`](evals/BENCHMARK.md) publishes the results: 79–80/80 routing accuracy on 12 of the 16 published model configurations (GPT-5.4/5.5/5.6 and Claude Sonnet/Opus/Fable families; the outliers are a documented strict-reading mode, not crashes), plus catalog-ablation runs showing the skills absorb each other's traffic when installed as a subset. Reproduce with `bun evals/run-eval.ts build`.\n\n## Manual binary install\n\nIf you need to install the native host without the CLI, download the binary for your platform from the [latest release](https://github.com/dundas/thinkrun/releases/latest) and run:\n\n```bash\nchmod +x thinkbrowse-host-* && ./thinkbrowse-host-* --install\n```\n\n> The native host binary keeps the `thinkbrowse-host` filename for now — it's the same signed binary, unchanged by the rename.\n\n## FAQ\n\n### How do I give Claude Code (or Cursor) access to my browser?\nAdd ThinkRun as an MCP server — `npx @thinkrun/mcp` — and your agent can drive and read a real Chrome session, your cookies and logins included. See \"Add to your MCP config\" above.\n\n### How do I show Cursor a UI bug instead of pasting a screenshot?\nRecord the bug with ThinkRun. Instead of a flat image, your agent gets a structured artifact — the steps you took, the console errors, the failing network requests, and the DOM state at the moment it broke — so it fixes the actual problem instead of guessing from a picture.\n\n### Is there an MCP server that uses my real Chrome cookies and sessions?\nYes. ThinkRun runs against your real Chrome via a native host, so authenticated pages just work — no headless re-login, no separate profile, no fighting your browser for the Chrome profile.\n\n### How do I record a browser session for an AI coding agent?\nRun `npx @thinkrun/mcp` (or `npm i -g @thinkrun/cli`) and record any session — or let the agent record what it built. ThinkRun turns it into an AI-ready report (clicks, console, network, narration, screenshots, all time-synced), available as a shareable link and as LLM-ready Markdown/JSON.\n\n### How do I give my AI coding agent context about what went wrong?\nHand it a ThinkRun recording. The structured \"what happened\" — actions, first bad signal, expected vs actual, and the element state — is exactly the context an agent needs to land a fix on the first try.\n\n## More\n\n- [thinkrun.ai](https://thinkrun.ai?utm_source=github&utm_medium=readme&utm_campaign=thinkrun-repo)\n- [Privacy Policy](https://thinkrun.ai/privacy)\n",
  "bytes": 5216,
  "sha": "eab972ec8df0d429eba81475346598d6fa9867d8565a10186e5b7cee60534fdc",
  "repo_slug": "dundas/thinkrun",
  "fonte": "repo",
  "truncated": false,
  "api": "https://agentalog.com/api/listings/mcp_io_github_dundas_thinkrun_9b47e99d/readme"
}