Content
71%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 well-structured, reasonably concise workflow with a concrete, verified script-call contract that matches the bundled generator. Its two real defects are a dangling reference to a nonexistent reference.md and the absence of any post-generation verification step; a fully worked payload example would further strengthen actionability.
Suggestions
Fix or remove the dangling `reference.md` pointer — either add the referenced file (layout/colors/python-pptx details) to the bundle or drop the section that tells the agent to read it.
Add a post-generation validation step, e.g. confirm the output .pptx exists and contains the expected 8-12 slides before reporting success.
Include one fully worked payload example (a complete JSON with title, subtitle, and 2-3 slides) so the contract is copy-paste ready, and deduplicate the 8-12-page / 3-5-bullet rules stated in both the flow and the constraints section.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean and free of concept explanations Claude already knows, but the 8-12 page count and 3-5 bullet rules are each stated twice (in the 规划页序 step and again in the 约束 section), so it sits at 'efficient; minor instances… that could be trimmed' rather than 'every token earns its place'. | 4 / 5 |
Actionability | It gives the exact tool call signature (run_skill_script(name="pptx", script="generate_pptx.py", payload=...)) and a concrete payload JSON schema whose fields match the bundled generator script. It stops short of a fully worked filled-in payload example, leaving it at 'mostly executable guidance with minor gaps' rather than copy-paste ready. | 4 / 5 |
Workflow Clarity | The four-step sequence (通读来源 → 规划页序 → 提炼要点 → 生成文件) is clear and enumerates the required slide structure explicitly. However there is no post-generation validation step (e.g., confirm the output file was created with the expected slide count), which is a minor checkpoint gap; the destructive/batch cap does not apply since this is a single generative, non-destructive script call. | 4 / 5 |
Progressive Disclosure | The layered structure is good in principle (core flow in SKILL.md, details deferred to `reference.md` and `scripts/generate_pptx.py`, one level deep), but the bundle actually contains only scripts/generate_pptx.py — `reference.md` does not exist, so the pointer is a dead reference that would fail when an agent follows it. Per the guideline to score against the actual bundle structure, this lands at 'some structure but could be better organized' with a broken reference. | 3 / 5 |
Total | 15 / 20 Passed |