jwbaldwin avatar

morning-report

Generate a morning MR status report. Use when the user says "morning report", "mr status", "what nee

提供方 jwbaldwin|开源

Morning Report

Generate a GitLab MR status report for James. This is an AI assistant giving a morning briefing - use natural language, not terse shorthand.

When to Use This Skill

Activate this skill when:

  • The user says "morning report" or "morning"
  • The user asks "what MRs need my attention?"
  • The user asks about their MR status
  • The user wants to know what to review

Workflow

1. Fetch My Open MRs

Use Zapier MCP to get all open MRs created by James:

zapier-mcp_gitlab_find_all_merge_requests({
  instructions: "Find all open merge requests that I created",
  scope: "created_by_me",
  state: "opened"
})

2. Fetch MRs Needing My Review

Use Zapier MCP raw API to get MRs where James is a reviewer:

zapier-mcp_gitlab_api_request_beta({
  instructions: "Get all open merge requests where I am a reviewer",
  method: "GET",
  url: "https://gitlab.com/api/v4/merge_requests?state=opened&reviewer_username=james.baldwin.z&scope=all"
})

3. For Each of My MRs, Gather Approval Status and Discussions

Fetch approval status and discussions in parallel for all of James's MRs:

zapier-mcp_gitlab_api_request_beta({
  instructions: "Get approval status for MR [iid] in project [project_id]",
  method: "GET",
  url: "https://gitlab.com/api/v4/projects/[project_id]/merge_requests/[iid]/approvals"
})

zapier-mcp_gitlab_api_request_beta({
  instructions: "Get discussions for MR [iid] in project [project_id]",
  method: "GET",
  url: "https://gitlab.com/api/v4/projects/[project_id]/merge_requests/[iid]/discussions"
})

4. Analyze Discussions for Feedback

For each of James's MRs, analyze the discussions to determine:

  1. Unresolved threads count - threads where resolvable === true && resolved === false
  2. Recent feedback - look for notes from other users (not James) in the last 24-48 hours
  3. Feedback summary - extract key points from recent comments

When summarizing feedback:

  • If there are 1-2 short comments, include the actual feedback text
  • If there are many comments or long discussions, summarize as "X unresolved threads to address"
  • Always attribute feedback to the reviewer (e.g., "Lucas asked about...")
  • Focus on actionable items and questions

5. Generate Report

Use the indented hierarchy format below. Write like an AI assistant giving a morning briefing - natural language, not terse keywords.

Output Rules

  1. Indented hierarchy, not tables - Tables don't render well in the terminal
  2. Natural language - Write "rebase with main" not "rebase". Write "From Lucas, opened 9 days ago" not "Lucas, 9d"
  3. Include the repo - Show which repo each MR is in (e.g., mcp or ai-command-center)
  4. Bold the MR titles - Use **[TICKET] title (!XXX)** format
  5. Truncate titles - If MR title is long, shorten it but keep the ticket ID
  6. Skip most labels - Only include labels that convey review urgency (like "review-in-depth")
  7. No branches - James knows his branches
  8. Relative time - "opened 9 days ago" not "2025-12-22"
  9. Stacked MRs - Note when MRs are stacked/dependent so James knows review order
  10. Links at the bottom - Keep the main report clean, put all URLs in a Links section at the end
  11. Summary line - End with a count of MRs waiting on action and MRs in review queue
  12. Greeting - Start with "Good morning, James" and the date
  13. Show reviewer feedback - For James's MRs with new comments or unresolved threads:
    • Show unresolved thread count in Status line
    • Add a "Feedback:" line with summary of what reviewers said
    • If feedback is short (1-2 sentences), quote it directly
    • If feedback is long or there are many threads, summarize (e.g., "3 unresolved threads - Lucas asked about error handling, Nate suggested refactoring")
    • Attribute feedback to the reviewer who left it

Status Determination Logic

For each MR, determine status by checking:

  1. Has Conflicts: has_conflicts === true or detailed_merge_status === "conflict" -> "has conflicts"
  2. Draft: draft === true -> "draft"
  3. Unresolved Discussions: Count threads where resolvable === true && resolved === false -> "X unresolved threads"
  4. Not Approved: Check approvals endpoint -> "X/Y approvals"
  5. Ready to Merge: detailed_merge_status === "mergeable" AND no unresolved threads -> "ready to merge"

IMPORTANT: An MR is NOT ready to merge if it has unresolved discussions, even if it has all required approvals.

Important Notes

  • For MRs to review, exclude any authored by James himself
  • Group stacked MRs together and note the dependency chain
  • If an MR is blocked by another MR, note that in the action
  • When an MR is fully approved but has conflicts, the action is "rebase with main, then ready to merge"
  • When an MR has unresolved discussions, it is NOT ready to merge even if fully approved
  • Only fetch discussions for James's MRs (not the review queue) to minimize API calls
  • When summarizing feedback, focus on questions and actionable items, skip "looks good" or approval-only comments
  • System notes (like "added X commits") are not feedback - filter these out

Example Output

Good morning, James. Here's your MR status for Monday, January 5th.

## Your MRs

**[AGP-1028] oxfmt (!229)** in `mcp`
  Status: has conflicts, 2/2 approved
  Action: rebase with main, then ready to merge

**[AGP-760] entitlements api (!225)** in `mcp`
  Status: 1/2 approvals, 2 unresolved threads
  Feedback: Lucas asked about the error handling in the feature flag check; Nate suggested adding a test for the edge case
  Action: address feedback, get second approval

**[AGP-1027] oxlint (!223)** in `mcp`
  Status: has conflicts, 0/2 approvals
  Action: rebase with main, request reviews

**[AGP-966] params fix (!4887)** in `ai-command-center`
  Status: 1/1 approved, 1 unresolved thread
  Feedback: Lucas asked "should we also handle the case where resolvedParams is empty?"
  Action: address feedback, then ready to merge

## To Review

**[AGP-973] redirect fix (!4922)** in `ai-command-center`
  From Lucas, opened 9 days ago

**[AGP-1031] MCP self-serve (!4913)** in `ai-command-center`
  From Dylan, opened 17 days ago
  Note: stacked on !4906, review that first

**[AGP-1003] agents self-serve (!4906)** in `ai-command-center`
  From Dylan, opened 17 days ago

**[AGP-953] hashedSecret verify (!224)** in `mcp`
  From Nate, opened 18 days ago
  Note: blocked by !4894, can't merge until that lands

**[AGP-1019] backfill hashed (!4894)** in `ai-command-center`
  From Nate, opened 18 days ago
  Note: stacked on !4890, review that first

**[AGP-952] longer secrets (!4890)** in `ai-command-center`
  From Nate, opened 19 days ago
  Label: review-in-depth

## Priority Actions

1. Rebase your 3 MCP repo MRs - all have conflicts with main
2. Review Nate's secret hashing stack starting at !4890, then !4894, then !224
3. Review Dylan's self-serve stack starting at !4906, then !4913

You have 4 MRs waiting on action and 6 MRs in your review queue.

---

**Links**

Your MRs:
- !229: https://gitlab.com/zapier/team-agents-platform/mcp/-/merge_requests/229
- !225: https://gitlab.com/zapier/team-agents-platform/mcp/-/merge_requests/225
- !223: https://gitlab.com/zapier/team-agents-platform/mcp/-/merge_requests/223
- !4887: https://gitlab.com/zapier/zapai/ai-command-center/-/merge_requests/4887

To Review:
- !4922: https://gitlab.com/zapier/zapai/ai-command-center/-/merge_requests/4922
- !4913: https://gitlab.com/zapier/zapai/ai-command-center/-/merge_requests/4913
- !4906: https://gitlab.com/zapier/zapai/ai-command-center/-/merge_requests/4906
- !224: https://gitlab.com/zapier/team-agents-platform/mcp/-/merge_requests/224
- !4894: https://gitlab.com/zapier/zapai/ai-command-center/-/merge_requests/4894
- !4890: https://gitlab.com/zapier/zapai/ai-command-center/-/merge_requests/4890

Future Consideration

GitLab CLI (glab) may be faster than Zapier MCP for fetching data. Worth exploring if report generation feels slow.

下载 Skill