Content
92%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 tightly structured, highly actionable scientific-writing workflow with explicit validation checkpoints, feedback loops, and clean one-level-deep references to a verified bundle. Its only real weakness is mild verbosity from inline time-sensitive dates and enumerative safety lists.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Largely operational and assumes Claude's competence with little concept-explanation padding, but inline time-sensitive markers ('researched on 2026-07-24', 'CONSORT 2025', 'ICMJE January 2026') and enumerative lists (the 'Never invent' bullets, intake list) could be trimmed; not a 5 because not every token earns its place, not a 3 since the COPE retirement is appropriately framed as deprecated rather than current. | 4 / 5 |
Actionability | Provides copy-paste-ready commands at each tooling step (scaffold, select_reporting_guidelines, check_consistency, validate_manifest, audit_claims, check_references, validate_authorship, lint_manuscript) with concrete arguments and file names, plus explicit ID conventions and inline claim markers, matching the fully-executable anchor. | 5 / 5 |
Workflow Clarity | A 12-step numbered workflow with explicit validation steps (validate_manifest, lint_manuscript, check_consistency), feedback loops ('Resolve every mismatch manually', 're-run every affected audit'), and human-approval gates/checklists (the declarations checklist, revision 7-step loop), matching the explicit-validation-with-feedback-loops anchor. | 5 / 5 |
Progressive Disclosure | SKILL.md is a clear overview with well-signaled one-level-deep references ('See references/...') and a complete 'Bundled files' inventory; all referenced scripts, references, and assets verified as real files, so navigation is easy and content is appropriately split. | 5 / 5 |
Total | 19 / 20 Passed |