MCP tool reference

Once you connect an AI agent to BugMojo over the Model Context Protocol, it gains a small set of tools for reading a bug's evidence and acting on it. This page lists each tool and what it does, so you know exactly what a connected agent can and cannot do.

These tools pull a bug and its captured evidence into the agent's session. They never change anything.

• list_bugs: List bugs you have access to, filtered by project, board, status, priority, label, or assignee. Returns titles, IDs, and current state so the agent can find the right bug.
• get_bug: Read one bug in full: title, description, repro steps, status, priority, labels, assignee, and device, OS, browser, and viewport metadata.
• get_replay_context: Read a structured summary of the rrweb session replay (the reconstructed DOM events and user actions leading to the failure). This is replay context for reasoning, not a video file.
• get_console_logs: Read the console logs captured during the session, including error messages and stack traces.
• get_network_requests: Read the network requests captured during the session, with their method, URL, status, timing, and masked headers. Bodies are never captured.

These tools change the bug. They run with your permissions, so an agent can only act where you already can.

• add_comment: Post a comment on a bug (for example, a summary of the root cause or a proposed fix) so it appears in the bug's activity for the whole team.
• update_bug_status: Move a bug between statuses: OPEN, IN_PROGRESS, DEV_TESTING, QA_TESTING, CLOSED, or REOPENED. Use it to reflect progress, but a human still verifies before a bug is truly done.

search_help lets the agent search this Help Center and read matching articles. It is how an agent answers questions about BugMojo itself (capture modes, statuses, or how a feature works) without you pasting docs by hand.