Content
53%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 lean and appropriately self-contained for a short skill, but its guidance is mostly abstract direction rather than executable instruction — no commands, code, or output template are provided. The workflow is inferable but not explicitly sequenced or validated.
Suggestions
Make the workflow an explicit numbered sequence (analyze request -> inspect related Python files -> prototype in temp scripts -> record findings -> clean up) with a validation checkpoint before writing the final research file.
Add concrete artifacts: an example temporary-test snippet and a short template showing the expected structure of the research .md file under .cursor/plans/{plan_name}/research.md.
Remove the duplicated first sentence and fix the 'Document base' heading to accurately describe the output-location section.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is short and directive with no explanations of concepts Claude already knows. It is not a 5 because the opening sentence duplicates the frontmatter description verbatim and several phrases ('Make reasonable technical decisions based on research', 'Use the python files with objective') could be tightened. | 4 / 5 |
Actionability | Most guidance is high-level direction ('Check python files related to the request', 'Identify what should be change', 'Document assumptions') with only one concrete artifact path (.cursor/plans/{plan_name_folder}/research as .md). There are no executable commands, code examples, or a template for the research file, leaving the specific steps to execute largely missing. | 2 / 5 |
Workflow Clarity | A rough sequence is discernible (analyze request -> prototype in temporary Python files -> decide from output -> document under .cursor/plans -> delete temp files), which goes beyond the bare 2-anchor list. However steps are unordered bullets, checkpoints are implicit ('based on the output decide what should be done'), and there is no verification of the final research document. | 3 / 5 |
Progressive Disclosure | The skill is under 50 lines, self-contained, and correctly collapses everything into one research file with no external references needed, satisfying the simple-skill exception. It is not a 5 because section organization is uneven — the oddly indented 'No user interaction required' block and the mislabeled 'Document base' heading create minor navigation gaps. | 4 / 5 |
Total | 13 / 20 Passed |