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.
A well-structured, actionable instruction-only skill with clear workflow, validation, and quality gates. The main gaps are minor redundancy between sections and tool references that assume rather than specify invocation.
Suggestions
Drop or condense the 触发条件 list in the body since the same triggers already appear in the frontmatter description, freeing tokens.
Clarify how to invoke the referenced skills (pitch-deck, ppt-maker, siliconflow-img-gen, etc.) — e.g., a one-line note on whether they are slash commands, MCP tools, or sub-skills.
Merge the 红线 (red lines) and 质量门控 (quality gates) into a single pre-delivery checklist to remove overlapping coverage of unverifiable claims and inconsistent numbers.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean — terse imperatives, tables, and checklists with no padding or explanation of concepts Claude already knows — but the trigger-conditions list repeats the description and the red-lines/quality-gate sections overlap slightly. | 4 / 5 |
Actionability | Guidance is concrete: a 12-slide deck structure, named tools (pitch-deck, ppt-maker, siliconflow-img-gen), and specific required financial-model components, but tool invocations reference other skills without confirming availability or invocation syntax. | 4 / 5 |
Workflow Clarity | A clear 5-step sequence includes a cross-check validation step, an explicit stop-and-confirm feedback loop on data conflicts, and a final quality-gate checklist, though per-step checkpoints are mostly implicit. | 4 / 5 |
Progressive Disclosure | The skill is appropriately self-contained with well-organized sections, tables, and checklists and no nested references, though at ~110 lines some material-type deep-dives could arguably live in separate files. | 4 / 5 |
Total | 16 / 20 Passed |