matthew-a-carr/babysit-pr
Drive a PR to merge: address review comments (human + Copilot), push fixes, wait for CI to go green, then squash-merge. Use when a human says "babysit PR #NNN", "address the comments and merge when green", or "get this PR landed". Pushes and merges — invoking it IS the authorization to do so. Refuses to merge on red CI, unresolved blocking reviews, or conflicts; escalates instead.
90
90%
Does it follow best practices?
Impact
94%
1.17xAverage score across 3 eval scenarios
Low
Low-risk findings worth noting
Evaluation results
87%
-4%Decide Whether to Merge Three PRs and Document the Reasoning
Merge prerequisite assessment with squash strategy and Dependabot triage
PR 301 merge decision
100%
100%
PR 301 squash strategy
100%
100%
PR 301 Conventional Commit title
100%
100%
PR 301 branch deletion
100%
100%
PR 301 label linkage noted
50%
100%
PR 302 blocked on CI
100%
100%
PR 302 blocker described
100%
100%
PR 303 Dependabot flag
66%
13%
No --admin flag mentioned
100%
100%
All 3 PRs covered
100%
100%
100%
5%Plan the PR Lifecycle Operations an Agent Runs
PR lifecycle script using mcp__github__* tool surface
mcp__github__ for PR read
100%
100%
mcp__github__ for comment reply
100%
100%
mcp__github__ for check status
100%
100%
mcp__github__ for merge
100%
100%
Squash merge strategy
100%
100%
Conventional Commit merge title
100%
100%
Branch deletion after merge
37%
100%
Merge guarded by green checks
100%
100%
No gh CLI for GitHub ops
100%
100%
97%
43%Prepare a Fix for a Reviewed Bug and Document the Verification Plan
TDD fix plan with full verification gate and Conventional Commit
Test written first
100%
100%
Test assertion described
100%
62%
Surgical change described
100%
100%
Full verification gate
33%
100%
OpenAPI check included
0%
100%
Conventional Commit format
100%
100%
No --no-verify
0%
100%
No test-skipping
0%
100%
No scope expansion
100%
100%
pnpm as package manager
0%
100%
- Evaluated
- Agent
- Claude Code
- Model
- Claude Sonnet 4.6