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 core sections are genuinely useful — accurate, executable examples that match the bundled script and a well-signaled one-level-deep API reference. The score is dragged down by substantial generic boilerplate (about half the document) and a broken validation example where the --help smoke test and expected output format do not match the script's actual behavior.
Suggestions
Delete or drastically compress the generic boilerplate sections (When Not to Use, Required Inputs, Recommended Workflow, Output Contract, Validation and Safety Rules, Failure Handling), which contain no bioRxiv-specific information.
Fix the Quick Validation section: replace "python scripts/biorxiv_search.py --help" with a working smoke test such as "python scripts/biorxiv_search.py CRISPR 7", and remove the fabricated expected-output block that the script never produces.
Ground the workflow in concrete commands — e.g. search, verify result count and DOI fields, then download PDFs and check the files are non-empty — instead of abstract directives like "select the documented execution path".
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Roughly half the body is generic template boilerplate — "Do not use this skill when the required source data, identifiers, files, or credentials are missing", "Keep the output safe, reproducible, and within the documented scope at all times", "Select the documented execution path and prefer the simplest supported command" — that teaches nothing specific to bioRxiv and pads the document. This matches anchor 2 ("several unnecessary explanations or padded sections"); it is not 3 because the padding is substantial rather than occasional, and not 1 because the core sections (When to Use, Example Usage, Implementation Details) are concrete and useful. | 2 / 5 |
Actionability | The Python and CLI examples are executable and match the actual bundled script ("searcher.search_by_keywords([\"CRISPR\"], days_back=30)", "searcher.download_pdf(first[\"doi\"], out_path)", "python scripts/biorxiv_search.py \"CRISPR\" 30"), with an install command included. It stops short of 5 because the "Quick Validation" path "python scripts/biorxiv_search.py --help" is broken (the script has no argument parsing and would treat --help as a search keyword) and the "Expected output format" block ("Result file: biorxiv_database_result.md") describes output the script does not produce. | 4 / 5 |
Workflow Clarity | A sequence is present ("Validate the request... Select the documented execution path... Produce the expected output... Run a final validation pass"), but the steps are abstract and the only concrete checkpoint (the --help smoke test) is incorrect, matching anchor 3's "sequence present but checkpoints missing or implicit". It is not 2 because the end-to-end Python example does show a concrete ordered flow (search, inspect metadata, download PDF), and not 4 because no working validation command is actually given. | 3 / 5 |
Progressive Disclosure | Structure is good: the body is organized into clear sections, the API detail is appropriately split into a real one-level-deep reference ("See references/api_reference.md for endpoint/field details and response formats" — the file exists and contains exactly that), and the script path scripts/biorxiv_search.py is real and correctly referenced. It falls short of 5 because the overview itself is diluted by inline boilerplate sections (Output Contract, Validation and Safety Rules, Failure Handling) that add navigational noise rather than pointing to purposeful content. | 4 / 5 |
Total | 13 / 20 Passed |