Content
77%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 well-structured, mostly lean instructional skill with a clear validated workflow, explicit refusal and recovery contracts, and concrete per-output requirements. Its main weaknesses are duplicated deliverable listings and, more importantly, three sizable bundled reference documents that are completely unlinked from SKILL.md.
Suggestions
Add pointer lines in the body (e.g., under 'Script Usage Notes' or 'Authorized Input Sources') referencing references/api_reference.md, references/authentication_guide.md, and references/integrations.md so the bundled reference material is discoverable.
Merge the 'Primary Writing Outputs' section with the 'Writing Output Contract' section to eliminate the duplicated enumeration of the five deliverables.
Include one worked example of a script invocation with real arguments (beyond '--help') and a brief filled-in template excerpt to make the writing outputs concrete.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is list-based and lean, assuming Claude's competence (no explanation of what an ELN or Methods section is), but the five deliverables are enumerated twice ('Primary Writing Outputs' then 'Writing Output Contract') and the opening framing ('Its goal is not just API access, but turning authorized ELN material into manuscript-ready writing deliverables') is padded. | 4 / 5 |
Actionability | Concrete script commands ('python scripts/setup_config.py', '--help', 'Use --dry-run where available before live execution'), an exact refusal template, and a named template asset ('assets/writing_outputs_template.md') give mostly executable guidance; minor gaps are the absence of a real-argument script invocation and any worked example of a writing output. | 4 / 5 |
Workflow Clarity | The five-step workflow has explicit validation checkpoints (step 1 authorization/sufficiency gate with a stop-and-refuse path, step 5 safety pass), feedback loops for error recovery (refusal contract and the 'If a script path fails' recovery instructions), and a completion checklist, matching the top anchor. | 5 / 5 |
Progressive Disclosure | SKILL.md appropriately points to the three bundled scripts and the template asset, but the three substantial files in references/ (api_reference.md, authentication_guide.md, integrations.md) are never mentioned anywhere in the body, so those references are present but not clearly signaled. | 3 / 5 |
Total | 16 / 20 Passed |