Content
75%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 an efficient, actionable ruleset with a concrete verification workflow and clear sectioning. It stays just below top marks because of the non-essential Provenance token cost and incompletely specified error-recovery loops.
Suggestions
Trim or collapse the Provenance UUID list (or move it to a reference file) to recover tokens that do not aid execution.
Add an explicit fix-and-retry loop for the failure patterns (e.g., when check:pack flags a private dep leak or lockfile drift, state the corrective next step) to push workflow clarity higher.
Move the Failure Patterns detail into a separate reference file and link to it from the rules, keeping SKILL.md as a tighter overview.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Lean rule statements with brief rationale assume Claude's competence and avoid explaining basics, though the Provenance UUID list and the dense rule 5 add some non-essential tokens. | 4 / 5 |
Actionability | Concrete commands and config keys ('pnpm install --frozen-lockfile', 'check:pack', 'build.rolldownOptions.external', 'externalizeInstallableDependencies') give mostly executable guidance with minor gaps. | 4 / 5 |
Workflow Clarity | Rule 6 sequences verification (check:pack, frozen-lockfile, lockfile-drift inspection, clean packed install) with most validation checkpoints present, but error-recovery feedback loops are only partly spelled out. | 4 / 5 |
Progressive Disclosure | Well-organized into Rules / Failure Patterns / Provenance with no nested references and no bundle files, though the single-file layout keeps some content inline that could be split out. | 4 / 5 |
Total | 16 / 20 Passed |