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.
Highly actionable, well-structured reference content: executable commands, exit codes, error taxonomy, validation checkpoints, and a dry-run gate for the batch migrate operation. The main costs are token inefficiency — a duplicated frontmatter template and dated observed-output transcripts — and inline bulk detail that a references/ bundle would offload.
Suggestions
Collapse the duplicated frontmatter documentation: the show-template observed output and the Frontmatter schema section show nearly identical YAML — keep one and cross-reference it from the other.
Trim or move dated observed-output transcripts (list output, fetch-github draft sample) into a references/ file, and keep version/date-stamped details out of the main body to avoid time-sensitivity rot.
Verify the out-of-bundle links (../research-curator/references/entry-template.md, ../../../plugins/development-harness/docs/sdlc-layers/) resolve from this skill's install location, or inline the one-line summary each link exists to provide.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is operational and free of concept explanations Claude already knows, but the full frontmatter template is printed twice (show-template output and the Frontmatter schema section), and dated observed-output transcripts ("Observed output (2026-02-22)", "verified 2026-02-24", versions like "v2.64.0") pad the file without being quarantined in a deprecated/old-patterns section. Fits "mostly efficient but includes some unnecessary explanation or could be tightened"; not 4 because the duplication and dated transcripts exceed minor trimming, not 2 because padding is confined to a few spots. | 3 / 5 |
Actionability | Every command has an executable `uv run` invocation, exit codes are documented ("0 = success, 1 = parse/write error, 2 = validation error"), each error type carries a fix ("Fix: gh auth login"), and the examples are copy-paste ready — the fetch → edit → add workflow and a full `set-description` command with a generated description string. The description rules are explicitly marked "pass verbatim to each subagent". Fully executable guidance covering the common cases. | 5 / 5 |
Workflow Clarity | Sequences include explicit validation checkpoints: fetch-github requires reviewing the draft before `add`; migrate is gated by "Run with --dry-run first to preview changes before committing" plus a Migrated/Failed summary; generate-descriptions has exit-code feedback ("The orchestrator only hears back from an agent on non-zero exit") and post-write validation via `uvx skilllint@latest check --fix`; update-append documents its abort conditions. The batch operations (migrate, add) all carry validation, so the batch-operation cap does not apply. | 5 / 5 |
Progressive Disclosure | A single-file skill with well-organized per-command sections and a useful source-line map, but the observed-output transcripts and duplicated frontmatter schema are inlined where reference files would serve, and links point outside the bundle to sibling skills ("../research-curator/references/entry-template.md", "../../../plugins/development-harness/docs/sdlc-layers/") that are not part of this skill's bundle. Good structure with minor organization gaps fits anchor 4; not 5 since this is a 350-line skill whose bulk detail could be split out, not 3 since sections are clear and well signaled. | 4 / 5 |
Total | 17 / 20 Passed |