For AI coding agents

Feed your agent the whole bug, not a screenshot

Coding agents fix the wrong thing when they only see a screenshot and a description. Brie exposes every captured bug through MCP and a REST API, so your agent can pull the console trace, replay the steps, and inspect the failing network call before it writes a single line of code.

Access every capture programmatically

The REST API returns the full slice payload, screenshots, annotations, session replay events, console logs, and network requests. Agents fetch by slice ID and get exactly what a human reviewer would see.

Native MCP support

Point Claude Code, Cursor, or any MCP-aware client at the Brie MCP server. Your agent can list open reports, load a specific bug, and ground its fix in real captured state, no copy-paste from the tracker.

Ground fixes in the actual repro

The agent scrubs through the session replay, reads the console error that fired, and inspects the network response body. It ships a fix that addresses the observed failure, not a hallucinated one.

From bug report to PR without a handoff

Wire Brie into your agent runner: new report lands, agent picks it up, opens a PR against the right repo with the fix and a link back to the source capture. Humans review, not transcribe.

Try it on your next bug

Install the Brie extension and capture your first report in under a minute.

Get started free