Content
77%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
This is a solid, actionable skill with clear workflow sequencing and good validation checkpoints throughout the merge process. Its main weakness is some redundancy between the Safety, Completion Criteria, and Guardrails sections, which inflates token usage without adding new information. The executable shell commands and explicit error handling paths make this highly practical for Claude to follow.
Suggestions
Consolidate the 'Safety', 'Completion Criteria', and 'Guardrails' sections into a single constraints section to eliminate redundancy and save tokens.
Remove the 'Execution Rule' meta-instruction about not stopping after printing the checklist — this is behavioral coaching that could be in the frontmatter description rather than consuming body tokens.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The skill is mostly efficient but has some redundancy — the 'Guardrails' section at the end largely repeats the 'Safety' section, and the 'Completion Criteria' section overlaps with the workflow steps. The 'Execution Rule' section telling Claude to actually execute commands is somewhat meta-unnecessary. However, most content is actionable and not padded with explanations of basic concepts. | 2 / 3 |
Actionability | The skill provides fully executable shell commands for every step, with concrete gh CLI invocations, git commands, and conditional logic. Commands are copy-paste ready with clear placeholders (<PR>), and error handling paths are specified (e.g., checks pending → use --auto, merge fails → stop and report). | 3 / 3 |
Workflow Clarity | The multi-step process is clearly sequenced with explicit validation checkpoints: sanity checks before merge (draft status, failing checks, behind main), state verification after merge (step 6), and conditional cleanup only after confirmed MERGED state. There are clear feedback loops — if checks fail, stop and run /preparepr; if merge fails, stop and report. | 3 / 3 |
Progressive Disclosure | The content is well-structured with clear sections and headers, but it's a fairly long monolithic file (~130 lines of content) with no references to external files for detailed information. Some sections like the guardrails/safety overlap could be consolidated. For a skill of this complexity, the inline approach is acceptable but not optimal. | 2 / 3 |
Total | 10 / 12 Passed |