Check MR Feedback

Review your open MRs for feedback requiring response.

Inputs

Workflow

1. Load Persona

• •persona_load("developer")

2. Check Known Issues

• •check_known_issues("gitlab_mr_list", "")

3. Get Open MRs

• •gitlab_mr_list(project, author="@me") — open MRs
• •Filter by mr_ids if provided

4. Get Comments (up to 5 MRs)

• •gitlab_mr_comments(project, mr_id) for each MR

5. Analyze Comments

• •Use split_mr_comments, is_bot_comment (shared parsers)
• •Filter: human comments only, exclude own
• •Detect: meeting requests (meeting, call, discuss, sync, chat, walk through)
• •Detect: needs response (please, could you, ?, consider, suggest)
• •Build: {mr_id, title, comments, meeting_requests, needs_response}

6. Meeting Invites (if create_meetings)

• •Check google_calendar config
• •For each meeting request: google_calendar_quick_meeting(title, attendee_email, when="auto")
• •Or suggest tool calls if not auto-creating

7. Format Summary

• •Per MR: author, text, icon (💬/📅/❓)
• •Meeting requests: suggest google_calendar_find_meeting, google_calendar_quick_meeting

8. Memory

• •memory_session_log("Checked MR feedback", "{count} MRs with feedback")
• •Add follow-ups: "Respond to feedback on MR !X", "Schedule meeting for MR !X with Y"

9. Learn from Failures

• •VPN: learn_tool_fix("gitlab_mr_list", "no such host", "VPN", "vpn_connect()")

MCP Tools

• •persona_load
• •check_known_issues
• •gitlab_mr_list
• •gitlab_mr_comments
• •google_calendar_quick_meeting (if create_meetings)
• •memory_session_log
• •memory_append (follow_ups)
• •learn_tool_fix

Output

• •MR Feedback Summary
• •Per MR: comments with author, text, meeting/response flags
• •Meeting scheduling suggestions
• •No pending feedback if clear