Content
100%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
A well-structured QC skill: lean, actionable, and properly decomposed across real bundle files. The single runnable command is verified executable, each QC dimension has concrete criteria or a checklist, and the four dimensions flow into a clear severity-based report with an explicit empty-result checkpoint.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Lean body that assumes Claude's competence — no explaining what a deck or library is — with brief, motivating asides ("where decks go wrong quietly") that illustrate failure modes rather than pad. Every section earns its tokens. | 3 / 3 |
Actionability | Provides a copy-paste-ready verified command ("python scripts/extract_numbers.py /tmp/deck_content.md --check"), a concrete input format spec, and concrete verification criteria (e.g. "#1 in a $100B market" with $200M revenue = 0.2% share), plus real referenced files for terminology and report structure. | 3 / 3 |
Workflow Clarity | Clear sequenced workflow (read deck → four numbered QC dimensions → severity-categorized output) with an automated check (the script flags conflicting values) and checklists per dimension; the empty-findings rule ("no number inconsistencies found" is a finding) acts as an explicit completeness checkpoint. | 3 / 3 |
Progressive Disclosure | Overview body points to well-signaled, one-level-deep, verified references (scripts/extract_numbers.py, references/ib-terminology.md, references/report-format.md); detailed terminology and the report template are appropriately split out rather than inlined. | 3 / 3 |
Total | 12 / 12 Passed |