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.
Excellent operational content: fully executable commands, explicit verification with feedback loops, and nuanced gh-specific gotchas. Minor room to trim duplicated routing guidance and shell explanations, and to split edge-case procedures into reference files.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense and gh-specific throughout (acceptance table, pairing semantics, \r handling, publish-no-undo warnings), but the Evidence-placement routing is stated twice (intro and "Description, for PR evidence") and it over-explains shell behavior Claude knows ("the quotes are what keep the shell from reading that `#` as a comment"). Not 5 because of these minor trims; not 3 because nearly all content is non-obvious and earns its place. | 4 / 5 |
Actionability | Every workflow ships copy-paste-ready commands: `gh pr edit <pr> --attach ...`, the awk placement pipeline, `printf ... | gh pr create --body-file -`, verification greps, and the `<img>` sizing snippet, covering comments, PR creation, existing descriptions, and multi-image cases. Not 4 because there are no gaps in executable detail for the common cases. | 5 / 5 |
Workflow Clarity | Clear sequence with explicit validation checkpoints and feedback loops: confirm files and destination before publishing, upload in one command, then Verify counts `user-attachments` and leftover local paths, with recovery paths (short count → re-attach only the missing; above zero → rewrite the body rather than uploading a duplicate) and a result-to-action table. The destructive/batch cap does not apply since validation is thorough. | 5 / 5 |
Progressive Disclosure | Sections are well-organized with clear cross-references ("**Verify** below", "**Place an image in an existing description**"), but this is a ~170-line self-contained document, above the under-50-line simple-skill exception, and the placement recipe and Sizing sections are peripheral procedures that could live in one-level-deep reference files. Not 5 because there is no content split at all; not 3 because what is inline is well-structured and nothing is buried. | 4 / 5 |
Total | 18 / 20 Passed |