Content
70%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 thorough, well-sequenced orchestration protocol with strong workflow clarity and executable examples. Its weaknesses are length/repetition and a monolithic structure that should offload reference material into bundle files.
Suggestions
Move the epistemic-instrumentation artifact schemas (intent-diff, claim-graph, observation-manifest field lists) and the Search craft operator table into reference files under ./references/, leaving SKILL.md a concise overview that links to them.
Collapse the Failure modes table into the phase sections it references, or move it to a reference file, to remove the duplicated emphasis that inflates token cost.
Tighten the repeated 'always/never' authority framing — state the override-once rule once in the Authority section and let the phase steps reference it without restating.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | It does not explain concepts Claude already knows, but it is padded with repeated rhetorical emphasis — the authority claims are restated, and the Failure modes table largely duplicates rules already stated in the phase sections — so it could be tightened considerably. | 3 / 5 |
Actionability | Provides concrete executable guidance — `task(subagent_type="explore", ...)` spawn blocks, the `mkdir -p .omo/ulw-research/...` command, `chrome --headless --print-to-pdf`, `pandoc ... -o <out.docx>`, and the EXPAND tail template — with only minor gaps from placeholder fields like `<angle>` and `<lead>`. | 4 / 5 |
Workflow Clarity | A clear Phase 0-5 sequence with explicit validation checkpoints (format-proposal gate, visual-QA delivery gate, asset-manifest verification) and feedback loops (validate -> fix -> re-render; convergence rules), so the destructive/batch cap does not apply. | 5 / 5 |
Progressive Disclosure | Section headers are well organized, but the skill is a ~394-line monolith with no bundle files; the epistemic-instrumentation field schemas, Search craft, and Failure modes clearly belong in separate reference files rather than inlined in SKILL.md. | 3 / 5 |
Total | 15 / 20 Passed |