Content
63%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 well-structured, actionable workflow with concrete gh CLI commands and sensible safeguards (confirmation gate, dry-run, idempotency, error handling). Main weaknesses are redundancy (the Task Execution section repeats earlier content), a missing post-action verification loop for batch GitHub operations, and no use of reference files to keep the 300-line body lean.
Suggestions
Delete the 'Task Execution' section: its argument-handling rules duplicate 'Usage' and 'Step 1'; fold any unique details (e.g., --auto-approve behavior) into those sections.
Move the ~35-line example triage summary and the .claude/pr-review-triage.local.md configuration spec into a references/ file (e.g., references/examples.md and references/config.md) and link to them from the body.
Add a post-action verification step for the batch operations: after acting, re-check that each reaction/reply/issue actually posted (e.g., re-fetch the thread or issue) and retry or report failures, closing the validate→fix→retry loop.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly efficient (tables, concrete commands, terse decision criteria), but the "Task Execution" section largely duplicates argument handling already covered in "Usage" and "Step 1", and the ~35-line sample summary block is padded relative to its instructional value. Fits 'mostly efficient but could be tightened' rather than 'efficient with only minor trims'. | 3 / 5 |
Actionability | Concrete, executable `gh` commands appear for every action (fetch, react, reply, resolve) plus a category table, decision framework, and formatted example output. Minor gaps: beads issue creation offers two paths ("Use the /beads:create skill or create issues directly") without specifics for the direct route, and graphql placeholders need assembly — so not fully copy-paste ready throughout. | 4 / 5 |
Workflow Clarity | A clear 5-step sequence with pre-action checkpoints: present summary → "Wait for user confirmation before proceeding", dry-run preview, idempotency checks before acting, and an error-handling section. Not a 5 because the batch GitHub actions lack a post-action verify/retry feedback loop (e.g., re-checking that a reaction or reply actually posted before marking a comment handled); not a 3 because validation before acting is explicitly present. | 4 / 5 |
Progressive Disclosure | Sections are well-organized with clear headers, but at ~300 lines with no reference files, the large example summary block and the configuration spec are inline content that could be offloaded to a references/ file to keep SKILL.md an overview. Fits 'some structure but could be better organized' — no nested-reference problem exists since there are no references at all. | 3 / 5 |
Total | 14 / 20 Passed |