A Better Way to Handoff Web Bugs to AI Agents

I’ve been experimenting with AI-assisted debugging and noticed a recurring gap: most tools optimize for agent-led exploration (ex: giving claude code a browser to click around and try to reproduce an issue). But in many cases, I've already found the bug myself. What I actually want is a way to hand the agent the exact context I just saw - without retyping steps, copying logs, or hoping it can reproduce the behavior.

So we built FlowLens, an open-source MCP server + Chrome extension that captures browser context and lets coding agents inspect it as structured, queryable data.

The extension can: - record specific workflows, or - run in a rolling “session replay” mode that keeps the last ~1 minute of DOM / network / console events in RAM.

If something breaks, you can grab the “instant replay” without reproducing anything. The extension exports a local .zip file containing the recorded session. The MCP server loads that file and exposes a set of tools the agent can use to explore it.

One thing we focused on is token efficiency. Instead of dumping raw logs into the context window, the agent starts with a summary (errors, failed requests, timestamps, etc.) and can drill down via tools like search_flow_events_with_regex or take_flow_screenshot_at_second. It can explore the session the way a developer would: searching, filtering, inspecting specific points in time.

Everything runs locally; the captured data stays on your machine.

Repo: https://github.com/magentic/flowlens-mcp-server