Content
88%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 tight, well-sequenced operational skill body with executable commands and genuine validation checkpoints. The main gaps are the absence of any concrete example for native (non-compat) operations and all content being inlined rather than split into reference files.
Suggestions
Add one minimal example request and status-poll for a native operation (e.g. creating a dubbing job and checking its progress), marked as shape-only since endpoints evolve.
Move the desktop-workflows section (Electron UI, dictation, dubbing export checks) into a references/ file to slim SKILL.md to core API workflows.
Include a short example for reference-based voice cloning (reference upload + profile creation) or link a dedicated guide, since that flow currently has no concrete command.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Lean and dense throughout — e.g. "The default backend is `http://localhost:3900`; honor the user's configured address" — with no tutorial padding, no explanations of concepts Claude already knows, and every sentence carrying project-specific non-obvious information. | 5 / 5 |
Actionability | Speech and transcription flows are fully executable (health/discovery GETs, copy-paste curl with JSON body, --fail-with-body, output probing), but cloning/design/dubbing sections defer entirely to runtime OpenAPI discovery with no example payloads or job-status calls — an explicitly justified gap that still leaves common native operations without concrete commands. | 4 / 5 |
Workflow Clarity | Numbered connect sequence with health/discovery checkpoints; generation flow requires checking HTTP status and probing output ("an error response can be written to the output path"); async jobs inspect terminal errors and the error-handling section enumerates failure classes with a fix-the-reported-condition feedback loop. | 5 / 5 |
Progressive Disclosure | No bundle files exist and the ~82-line body is entirely inline with clear section headers; well-structured and navigable, but material like the desktop-workflows section could be split into a one-level-deep reference file, so it falls short of the clear-overview-with-references anchor. | 4 / 5 |
Total | 18 / 20 Passed |