Content
81%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 highly actionable and workflow-safe: every step has executable commands, explicit validation gates, fallbacks, and error recovery, which is exactly what a risky write-operation skill needs. The main weakness is redundancy — the same safety rules are repeated across three dedicated sections plus troubleshooting — which inflates token cost without adding new information.
Suggestions
Consolidate the "Critical Guidelines", "Red Flags", and "Rationalization Table" sections into one guardrail section (e.g. a single excuse-vs-reality table); the current three formats restate the same rules nearly verbatim.
Remove filler and typos: the "If you fail the task, you will be killed!" threat adds no operational value, and "or simular command" should be "similar".
Consider moving the Parameters Explained table and the rationalization examples into a reference file (e.g. references/guardrails.md) to keep SKILL.md as a lean overview, since no bundle files currently exist.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The operational Steps 0–5 are tight and concrete, but the same guardrails are restated three times in different formats ("Critical Guidelines" bullets, "Red Flags" bullets, the "Rationalization Table") and repeated again in Troubleshooting rows, plus filler like "If you fail the task, you will be killed!". Not 2 because the majority of the body is efficient, executable instruction; not 4 because the guardrail duplication is genuine padding by token-efficiency standards. | 3 / 5 |
Actionability | Fully executable throughout: `gh auth status`, `gh pr view --json number,url,headRefName`, a complete GraphQL query with the jq filter, the `resolveReviewThread` mutation, MCP invocation blocks with concrete parameters, a parameter table, and error-message-specific troubleshooting entries. Copy-paste ready and covering the common cases. | 5 / 5 |
Workflow Clarity | Clear Steps 0–5 with a disposition decision table and explicit validation checkpoints ("No verification → no resolve", confirm `isResolved` is `true`, leave-unresolved-when-unsure default, fallback paths when a tool is unavailable, and wrong-thread recovery guidance). Although this is a batch operation, validation is thoroughly present, so the workflow-clarity cap does not apply. | 5 / 5 |
Progressive Disclosure | No bundle files exist; the skill is a single, well-sectioned SKILL.md with clear headers and easy navigation. Not 5 because at ~248 lines, content such as the Rationalization Table and the Parameters Explained reference table is inlined where a separate reference file would keep the overview leaner; not 3 because structure and placement are otherwise good. | 4 / 5 |
Total | 17 / 20 Passed |