Content
42%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 has a clear workflow skeleton and points to specific scripts and references, but it is weighed down by generic boilerplate and relies on a referenced prompts file that is absent from the bundle, leaving the actual screening logic unspecified. Validation is also only optional, weakening the workflow.
Suggestions
Remove or condense the generic template sections ('When to Use', 'Key Features', 'Implementation Details', 'When Not to Use', 'Required Inputs', 'Output Contract', 'Validation and Safety Rules', 'Failure Handling', 'Input Validation') and keep only abstract-screening-specific guidance.
Either bundle 'references/screening_prompts.md' or inline the actual screening role/logic and a concrete example input abstract with its Yes/No/Maybe decision and 'Result'/'Reason' JSON output.
Make the validation step a required checkpoint with a fix-and-retry feedback loop rather than marking it 'Optional', since screening is typically run over batches of papers.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is noticeably padded with generic boilerplate sections ('Use this skill when the request matches its documented task boundary', 'Execution model: validate the request, choose the packaged workflow') that restate process principles Claude already knows rather than skill-specific guidance. | 2 / 5 |
Actionability | It provides some concrete guidance (specific commands like 'python scripts/screen_paper.py --help', an output schema of strictly 'Result' and 'Reason', and a 3-step workflow), but the core screening logic is delegated to 'references/screening_prompts.md', which is not present in the bundle, leaving key details missing. | 3 / 5 |
Workflow Clarity | The Analysis/Formatting/Validation steps are sequenced, but validation is marked 'Optional' rather than an explicit checkpoint and there is no error-recovery feedback loop, so checkpoints are effectively missing. | 3 / 5 |
Progressive Disclosure | There is section structure and one-level-deep signaled references ('references/screening_prompts.md', 'scripts/screen_paper.py'), but those referenced files do not exist in the bundle and substantial generic content is inlined rather than split out. | 3 / 5 |
Total | 11 / 20 Passed |