perf(workflow): replace Copilot scripts with GitHub MCP server by timothyfroehlich · Pull Request #1096 · timothyfroehlich/PinPoint

timothyfroehlich · 2026-03-09T14:48:27Z

Summary

Deleted 4 shell scripts (585 lines) that burned excessive GitHub API calls — replaced by GitHub MCP server tools (pull_request_read, add_reply_to_pull_request_comment)
Removed review-polling from monitor-gh-actions.sh — the 20s poll loop was burning ~600 REST calls per CI run and causing 403 rate-limit errors
Consolidated label-ready.sh from 2 API calls to 1 using gh pr view --json statusCheckRollup
Created resolve-thread.sh — thin GraphQL stopgap for the one MCP gap (resolveReviewThread, upstream PR #1919 pending)
Updated 8 skill/doc files to describe MCP-first workflow with lightweight subagent (Haiku/Flash) for fetching review comments

Addresses bead PinPoint-e0e1.

Test plan

Verify pnpm run check passes (727 tests, all linters)
Verify resolve-thread.sh resolves a real thread on a test PR
Verify label-ready.sh correctly reads statusCheckRollup on a real PR
Verify monitor-gh-actions.sh still monitors CI without the review-polling subshell
Test MCP pull_request_read(method: "get_review_comments") returns Copilot threads
Test MCP add_reply_to_pull_request_comment posts replies to threads

🤖 Generated with Claude Code

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

The 20s review-polling subshell burned ~600 REST calls per CI run and was the root cause of 403 rate-limit errors. Review detection now happens via MCP after CI completes. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Use gh pr view --json statusCheckRollup to get CI status in the same call as PR metadata, eliminating a separate gh pr checks call. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Removed: - copilot-comments.sh (5 API calls) → MCP pull_request_read get_review_comments - respond-to-copilot.sh (3 API calls) → MCP add_reply_to_pull_request_comment + resolve-thread.sh - resolve-copilot-threads.sh (2+N calls) → MCP listing + resolve-thread.sh - pr-dashboard.sh (3×N calls) → MCP tools / gh pr list orchestration-status.sh updated to use gh pr list instead of pr-dashboard.sh. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

- Copilot comments fetched via MCP pull_request_read, not shell scripts - Replies via MCP add_reply_to_pull_request_comment - Resolves via resolve-thread.sh (MCP gap stopgap) - Lightweight subagent (Haiku/Flash) for large response protection Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Review detection now happens via MCP after CI completes, not during CI. References pinpoint-ready-to-review skill for the Copilot workflow. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Shell scripts replaced by MCP tools: - pull_request_read(get_review_comments) for fetching - add_reply_to_pull_request_comment for replies - resolve-thread.sh for the one remaining MCP gap Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Updated 5 files that still referenced copilot-comments.sh, respond-to-copilot.sh, resolve-copilot-threads.sh, and pr-dashboard.sh to use MCP tools and resolve-thread.sh instead. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

…write Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

vercel · 2026-03-09T14:48:41Z

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
pin-point	Ready	Preview, Comment	Mar 12, 2026 2:42am

supabase · 2026-03-09T14:48:43Z

Updates to Preview Branch (worktree/github-efficiency) ↗︎

Deployments	Status	Updated
Database	✅	Thu, 12 Mar 2026 02:41:55 UTC
Services	✅	Thu, 12 Mar 2026 02:41:55 UTC
APIs	✅	Thu, 12 Mar 2026 02:41:55 UTC

Tasks are run on every commit but only new migration files are pushed.
Close and reopen this PR if you want to apply changes from existing seed or migration files.

Tasks	Status	Updated
Configurations	✅	Thu, 12 Mar 2026 02:41:55 UTC
Migrations	✅	Thu, 12 Mar 2026 02:41:55 UTC
Seeding	✅	Thu, 12 Mar 2026 02:41:55 UTC
Edge Functions	✅	Thu, 12 Mar 2026 02:41:55 UTC

View logs for this Workflow Run ↗︎.
Learn more about Supabase for Git ↗︎.

Copilot

Copilot encountered an error and was unable to review this pull request. You can try again by re-requesting a review.

GitHub's statusCheckRollup sometimes includes entries with null name/status from third-party integrations (e.g. Supabase branching). The jq startswith() function crashes on null input. Filter out null-name entries early and fix parenthesization of the codecov exclusion filter. Found during testing PR #1096 — the Supabase Preview integration creates a ghost check entry with all-null fields. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Take our MCP-first language for Copilot review management with main's updated skill paths (.claude/skills/ → .agent/skills/). Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Copilot

Pull request overview

Copilot reviewed 18 out of 18 changed files in this pull request and generated 4 comments.

You can also share your feedback on Copilot code review. Take the survey.

Copilot · 2026-03-12T02:46:52Z

.agent/skills/pinpoint-ready-to-review/SKILL.md

+> 1. Call `pull_request_read` with `method: "get_review_comments"`, `owner: "timothyfroehlich"`, `repo: "PinPoint"`, `pullNumber: <NUMBER>`.
+> 2. Filter to threads where:
+>    - `isResolved` is `false`
+>    - The first comment's author login contains `copilot`


The subagent instructions say to filter review threads where the first comment’s author login “contains copilot”. That heuristic can match unrelated users/bots and could lead to replying to/resolving the wrong threads. Use an explicit allowlist of Copilot reviewer accounts (e.g., the same logins used by label-ready.sh: copilot-pull-request-reviewer / copilot-pull-request-reviewer[bot]), or otherwise match on the specific Copilot app/bot identity returned by MCP.

Suggested change

> - The first comment's author login contains `copilot`

> - The first comment's author login is exactly one of: `copilot-pull-request-reviewer`, `copilot-pull-request-reviewer[bot]`

Copilot · 2026-03-12T02:46:53Z

.agent/skills/pinpoint-commit/SKILL.md

+2. Filter to unresolved threads where the first comment author contains `copilot`
+3. For each comment, fix the code, then reply via MCP and resolve:


This step recommends filtering threads by whether the first comment author login “contains copilot”. That can accidentally include non-Copilot commenters (e.g., users with “copilot” in their username) and cause incorrect replies/resolutions. Prefer an explicit allowlist of the actual Copilot reviewer logins (as used elsewhere in repo scripts) or match on the Copilot app/bot identity returned by MCP.

Suggested change

2. Filter to unresolved threads where the first comment author contains `copilot`

3. For each comment, fix the code, then reply via MCP and resolve:

2. Filter to unresolved threads where the first comment author is identified as Copilot (for example, by matching the Copilot app/bot identity from MCP or using the explicit Copilot reviewer login allowlist used in repo scripts)

3. For each Copilot comment, fix the code, then reply via MCP and resolve:

Copilot · 2026-03-12T02:46:53Z

docs/plans/2026-03-09-github-efficiency-plan.md

+> 1. Call `pull_request_read` with `method: "get_review_comments"`, `owner: "timothyfroehlich"`, `repo: "PinPoint"`, `pullNumber: <NUMBER>`.
+> 2. Filter to threads where:
+>    - `isResolved` is `false`
+>    - The first comment's author login contains `copilot`


In the subagent prompt, filtering by “author login contains copilot” is overly broad and can match non-Copilot accounts. To avoid replying/resolving the wrong threads, tighten this to an explicit allowlist of the Copilot reviewer logins used by the repo’s scripts (e.g., copilot-pull-request-reviewer / copilot-pull-request-reviewer[bot]) or the specific Copilot app/bot identity returned by MCP.

Suggested change

> - The first comment's author login contains `copilot`

> - The first comment's author login is one of: `copilot-pull-request-reviewer`, `copilot-pull-request-reviewer[bot]`

Copilot · 2026-03-12T02:46:53Z

scripts/workflow/monitor-gh-actions.sh

-    ) &
-    REVIEW_WATCHER_PID=$!
-fi
+trap 'kill "${PIDS[@]}" 2>/dev/null || true; exit 0' TERM


The TERM trap exits with status 0. If this script is terminated (e.g., by a supervisor/timeout), callers that interpret 0 as “all passed” can get a false positive (the skills/docs explicitly treat exit code 0 as pass). Consider exiting with a non-zero status on TERM (e.g., 130/143) and/or writing a distinct signal file so aborted runs aren’t indistinguishable from success.

Suggested change

trap 'kill "${PIDS[@]}" 2>/dev/null || true; exit 0' TERM

trap 'kill "${PIDS[@]}" 2>/dev/null || true; touch "$SIGNAL"; exit 143' TERM

timothyfroehlich and others added 9 commits March 9, 2026 09:40

feat(workflow): add resolve-thread.sh — thin GraphQL stopgap for MCP gap

4a847e9

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

perf(workflow): consolidate label-ready.sh from 2 API calls to 1

08c4b22

Use gh pr view --json statusCheckRollup to get CI status in the same call as PR metadata, eliminating a separate gh pr checks call. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

docs(skill): simplify pinpoint-github-monitor, remove review-polling

3fd13e3

Review detection now happens via MCP after CI completes, not during CI. References pinpoint-ready-to-review skill for the Copilot workflow. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

docs: add design doc and implementation plan for GitHub efficiency re…

4fad400

…write Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Copilot AI review requested due to automatic review settings March 9, 2026 14:48

Copilot started reviewing on behalf of timothyfroehlich March 9, 2026 14:49 View session

vercel bot deployed to Preview March 9, 2026 14:50 View deployment

Copilot AI reviewed Mar 9, 2026

View reviewed changes

timothyfroehlich added the ready-for-review PR passed CI and has no unresolved review comments label Mar 11, 2026

vercel bot deployed to Preview March 11, 2026 02:38 View deployment

merge: resolve conflicts with main (skill paths + MCP-first wording)

48f8549

Take our MCP-first language for Copilot review management with main's updated skill paths (.claude/skills/ → .agent/skills/). Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Copilot AI review requested due to automatic review settings March 12, 2026 02:41

Copilot started reviewing on behalf of timothyfroehlich March 12, 2026 02:41 View session

vercel bot deployed to Preview March 12, 2026 02:42 View deployment

Copilot AI reviewed Mar 12, 2026

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

perf(workflow): replace Copilot scripts with GitHub MCP server#1096

perf(workflow): replace Copilot scripts with GitHub MCP server#1096
timothyfroehlich wants to merge 11 commits intomainfrom
worktree/github-efficiency

timothyfroehlich commented Mar 9, 2026

Uh oh!

vercel bot commented Mar 9, 2026 •

edited

Loading

Uh oh!

supabase bot commented Mar 9, 2026 •

edited

Loading

Uh oh!

Copilot AI left a comment

Uh oh!

Copilot AI left a comment

Uh oh!

Copilot AI Mar 12, 2026

Uh oh!

Copilot AI Mar 12, 2026

Uh oh!

Copilot AI Mar 12, 2026

Uh oh!

Copilot AI Mar 12, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

	> - The first comment's author login contains `copilot`
	> - The first comment's author login is exactly one of: `copilot-pull-request-reviewer`, `copilot-pull-request-reviewer[bot]`

		2. Filter to unresolved threads where the first comment author contains `copilot`
		3. For each comment, fix the code, then reply via MCP and resolve:

	trap 'kill "${PIDS[@]}" 2>/dev/null \|\| true; exit 0' TERM
	trap 'kill "${PIDS[@]}" 2>/dev/null \|\| true; touch "$SIGNAL"; exit 143' TERM

Conversation

timothyfroehlich commented Mar 9, 2026

Summary

Test plan

Uh oh!

vercel bot commented Mar 9, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

supabase bot commented Mar 9, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Copilot AI Mar 12, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 12, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 12, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 12, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

vercel bot commented Mar 9, 2026 •

edited

Loading

supabase bot commented Mar 9, 2026 •

edited

Loading