Content
72%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 an exemplary lean, imperative process guide with correct progressive disclosure to one real optional reference. Its weakness is actionability and workflow clarity: key operational concepts (Quick/Guided/Deep modes) and the output artifact are referenced without definition or example, leaving execution details implicit.
Suggestions
Define or point to where Quick, Guided, or Deep mode are specified (e.g. 'See the workflow manifest for mode definitions') instead of assuming external context the reader may not have.
Add a brief skeleton of the research document's expected sections or a one-line example finding entry (claim, source URL, date, confidence) to make the output artifact concrete.
Insert an explicit checkpoint before finishing, e.g. 'Confirm any decision-critical unknowns with the user before proceeding to the next stage,' to close the validation gap in the workflow.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean procedural instruction with zero concept explanation or padding — every sentence directs behavior ('Identify the decision the research must inform', 'Preserve unknowns explicitly and label assumptions'). It assumes Claude's competence and every paragraph earns its place, matching the 'lean and efficient' anchor. | 5 / 5 |
Actionability | There are concrete specifics (output path 'docs/research-[AppName].md', an explicit Handoff Context field list, source-verification requirements), but much of the guidance is abstract process direction with no example or template, and 'Use Quick, Guided, or Deep mode from the existing context' references modes undefined in this skill. This lands at 'some concrete guidance but incomplete', below the 'mostly executable' anchor 4 but above 'minimal concrete guidance' (2). | 3 / 5 |
Workflow Clarity | The paragraphs form a clear lifecycle sequence (identify decision, reuse context, ask questions, pick mode, investigate, record sources, save findings, hand off), but validation checkpoints are implicit and the Quick/Guided/Deep modes are externally defined without a pointer. This matches 'steps listed but validation gaps; checkpoints missing or implicit'; the destructive/batch cap does not apply since the skill performs no destructive operations, but there is not enough explicit validation to reach anchor 4. | 3 / 5 |
Progressive Disclosure | The body is well under 50 lines with a single, clearly signaled, one-level-deep reference ('consult the relevant [optional question prompts](references/question-bank.md)') that exists on disk and is explicitly marked optional. Content is appropriately split with nothing inlined that belongs in a separate file, matching the clear-overview anchor. | 5 / 5 |
Total | 16 / 20 Passed |