Content
92%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.
A strong, highly actionable skill body: a five-step audited workflow with a hard verification gate, complete executable code for the core layout discipline, and clean progressive disclosure into real, well-gated reference files. The only weakness is mild verbosity in the two motivational "why" sections, which read as design-rationale prose that could be compressed or moved into the references.
Suggestions
Compress "Why this is hard (and why a skill helps)" to the six-item drift-mode taxonomy alone — the one-line layout-discipline framing is already implicit in Step 4's rails.
Move the "Why geometry-based verification, not visual diff" section into references/layout-discipline.md (or a short note there), keeping only the one-line takeaway "geometry checks need only python-pptx and catch 1–2 mm overflows the eye misses" in SKILL.md.
The six drift modes in the taxonomy are re-explained across Step 4 and the anti-patterns section; state each mode once and cross-reference the fix step (e.g. "mode 3 → see the italic-preservation snippet") to cut ~15 lines.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense and action-oriented — rail constants, the Cursor class, severity rubric, and anti-patterns are all non-generic, project-specific knowledge Claude cannot be assumed to have. However, the "Why this is hard" preamble and the closing "Why geometry-based verification, not visual diff" section are discursive rationale, and the six-drift-mode taxonomy partially duplicates material elaborated in the references. Fits anchor 4 (efficient, minor over-explanation that could be trimmed) rather than 5, because two sections explain motivation at length where a tighter statement or a pointer to the references would do. | 4 / 5 |
Actionability | Every step gives copy-paste-ready, executable guidance: `scripts/extract_pptx.py <path-to.pptx> > pptx_dump.json`, complete runnable Python for the Cursor class, hero budget centering, and add_run italic preservation, an exact audit-table format with a severity rubric, and `scripts/verify_layout.py` with concrete override flags (--content-max-y / --canvas-h / --canvas-w). Matches the score-5 anchor — fully executable code covering the common cases; it is not 4 because there are no gaps requiring the reader to fill in details. | 5 / 5 |
Workflow Clarity | The five-step workflow is explicitly sequenced with a hard validation gate ("Zero violations is the gate for 'this re-export is shippable'. Don't claim the audit is fixed without running the verifier"), severity-driven fix priority, a mandated root-cause section, anti-pattern feedback ("a fix-without-audit pass tends to leave half the issues alive"), and a defined output checklist. This matches the score-5 anchor: clear sequence, explicit validation, and feedback loops for error recovery. | 5 / 5 |
Progressive Disclosure | The body is an overview plus the load-bearing summary of the technique; deep material (full layout rule set, five-layer font audit, table template) is split into one-level-deep references, each clearly signaled ("See `references/layout-discipline.md` for the full rules", "read `references/font-discipline.md") and gated by an explicit "Read the references when:" condition list. All five referenced bundle files exist and their contents match their descriptions. Matches the score-5 anchor — clear overview, well-signaled one-level-deep references, easy navigation. | 5 / 5 |
Total | 19 / 20 Passed |