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.
The body is an exemplar of actionable operator guidance: exact gh commands, explicit boarding criteria, validation checkpoints with error-recovery branches at every stage, and a concrete output format. Its only weaknesses are mild — some repeated safety assurances inflate the rationale section, and the whole procedure lives in one file with no bundle split. Anti-drift checks against neighboring anchors confirmed each score.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is command-first and tight, and its explanatory passages (invariant §7, why local rebases strip signatures) are genuinely project-specific rather than concepts Claude already knows. It falls short of the lean anchor at 5 because the not-bypassed/branch-protection assurances repeat between the intro and hard rule 4, and the invariant discussion runs three paragraphs that could be trimmed — fitting 'minor instances of over-explanation' at 4. | 4 / 5 |
Actionability | The procedure gives copy-paste-ready commands with exact flags and JSON fields (gh pr list --state open --json number,title,isDraft,…; gh api -X PUT "repos/{owner}/{repo}/pulls/$PR/update-branch"; gh run watch <databaseId>), concrete drop criteria for candidates, and a filled-in output template. The only deferral (merge-method flag) is explicitly justified by pointing to the authority document, so it matches the fully-executable anchor at 5 rather than 'minor gaps' at 4. | 5 / 5 |
Workflow Clarity | The five-step sequence has explicit validation checkpoints at every stage — green-check verification before boarding, `gh pr checks --watch` after update-branch, waiting on post-merge workflows between merges — plus feedback loops for error recovery (red after update → skip and record for /pr-iterate; failed post-merge workflow → hard stop and escalate). The batch-operation validation cap does not apply because validation is thorough, matching the anchor at 5 with recovery paths rather than 'most checkpoints' at 4. | 5 / 5 |
Progressive Disclosure | Sections are well organized (What this does, Hard rules, Procedure, Output format, Cross-references) and the external references (constitution/local-workflow.md, docs/adr/INDEX.md) are clearly signaled in a dedicated section, one level deep. At ~140 lines everything is inlined in one file — the rationale/governance discussion is long enough that a split or tighter overview could apply, and the under-50-line simple-skill exception does not apply, so 'good structure; minor organization gaps' at 4 is the best fit. | 4 / 5 |
Total | 18 / 20 Passed |