FlowLens MCP Server is an open-source tool designed to give AI-powered coding agents (like Claude Code, Cursor, GitHub Copilot / Codex, and others) full, replayable browser context to dramatically improve debugging, bug reporting, and regression testing for web applications. It works together with a companion browser extension: when a user reproduces a bug or a complicated UI interaction, the extension captures a rich session log, including screen/video recording, network traffic, console logs, DOM events, storage changes, and more, and exports it. The MCP server then loads this captured “flow” and exposes it to the AI agent via the Model Context Protocol (MCP), letting the agent examine, search, filter, and reason about the session just as a human developer would, without needing the agent to re-run the flow or rely on minimal reproduction data (logs, screenshots).
Features
- Integration with a browser extension to record comprehensive user flows: screen/video, network requests, console logs, DOM events, storage and navigation history
- Exposes recorded flows via MCP (Model Context Protocol) so AI coding agents can access full context for debugging, analysis, or test generation
- Supports installation via standard Python tooling (pipx install flowlens-mcp-server) for easy local setup
- Enables token-efficient bug reports: agent gets a structured summary and can drill down into relevant events (e.g. “errors”, “failed requests”, timestamps) rather than entire raw logs
- Useful for multiple use cases: bug reporting, reproducible debugging, regression tests, and even automated generation of test scripts (e.g. Playwright) from flows
- Open-source under Apache-2.0 license — transparent, modifiable, and usable as a baseline for building custom AI-driven debugging or QA tools