Content
77%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 a highly actionable, well-sequenced governed-CLI workflow with strong validation and feedback loops, but it is verbose in places and packs operational detail inline rather than splitting it into one-level reference files.
Suggestions
Deduplicate the awaiting_sources / no-op behavior: state the rule once and reference it from the upload, conflict, and recovery sections instead of re-explaining it three times.
Move the long update-check freshness and error-handling matrix (the "Keep the CLI and this Skill current" section) into a separate reference file and keep SKILL.md as a quick-start overview with a clear pointer.
Consolidate the scattered safety prohibitions (sudo/PATH/credential handling) into one short list rather than restating them across the bootstrap, auth, and source-upload sections.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The guidance is operational and assumes Claude's competence, but it restates the same invariants several times — the awaiting_sources no-op behavior is explained in three places (the source-upload section, the conflict section, and finish/recover), and the safety prohibitions are repeated across sections — so it is mostly efficient but could be tightened. | 3 / 5 |
Actionability | Fully executable, copy-paste-ready commands with exact flags are given throughout — `personwise version --json`, `course readiness --json`, `course create --input <blueprint.json> --json`, `source add/status/retry/detach`, `run wait/get/advance` — covering the common cases concretely. | 5 / 5 |
Workflow Clarity | A clearly sequenced workflow (version → bootstrap → auth → update check → readiness → blueprint → sources → checkpoints → publish) with explicit validation checkpoints and feedback loops, e.g. "Do not call run advance while any source is pending", CONFLICT→`read_current_state` handling, and retry-only-when-freshly-allowed recovery for a long-running batch operation. | 5 / 5 |
Progressive Disclosure | Referenced bundle paths are real one-level-deep files (`assets/bootstrap.sh`, `assets/bootstrap.ps1`, both present), but the SKILL.md itself is a ~330-line monolith that inlines operational detail which could live in separate reference files, with no quick-start-then-reference split. | 3 / 5 |
Total | 16 / 20 Passed |