Content
88%Weight 40%Scale 1-5Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
A tight, highly actionable PR-review workflow with concrete gh/git/python/bun commands, a well-sequenced six-step process, and explicit validation feedback loops for the risky force-push step. It appropriately defers to repository source-of-truth files rather than restating policy from memory.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Lean and assumes Claude's competence (no explanation of what a PR or gh is), with concrete commands throughout; a few prose guidance lines and the long flag list could be marginally tightened. Not a 5 because some explanatory sentences add length beyond the bare executable essentials. | 4 / 5 |
Actionability | Provides copy-paste-ready commands: a complete 'gh pr view ... --json' field list, 'gh pr diff', 'gh pr checks', 'git fetch origin refs/pull/<number>/head', 'python3 scripts/pattern_validator.py', and 'bun run build:data'. Fully executable guidance covering the common cases. | 5 / 5 |
Workflow Clarity | A clear six-step sequence (Inspect -> Validate -> Apply policy -> Run scripts -> Repair -> Produce review) with explicit validation checkpoints and feedback loops: re-check 'gh pr view' after push because mergeability may transiently show UNKNOWN, and use '--force-with-lease' for the destructive force-push to a contributor branch. | 5 / 5 |
Progressive Disclosure | Well-organized into labeled sections and correctly defers detail to repo files ('Treat CONTRIBUTING.md as the source of truth', 'TEMPLATE.md as the source of truth', scripts in scripts/). Not a 5 because some policy criteria in section 3 are inlined rather than purely pointing to CONTRIBUTING.md, leaving minor organization overlap. | 4 / 5 |
Total | 18 / 20 Passed |