Content
67%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 CLI skill body with executable commands, complete argument/output/error tables, and exemplary external-file navigation. It is held back by duplicated scope sections, a development-facing checklist, and multiple references to a tests/ bundle (data and test scripts) that is absent from the provided files.
Suggestions
Remove the missing tests/ references or ship the files: the "When to Read External Files" table, the Testing section, and the data provenance note all point to tests/data/, tests/run_tests.R, and tests/test_skill.R, none of which exist in the bundle.
Merge the duplicated out-of-scope guidance: "When Not to Use" and "Do not use this workflow for" under Input Validation repeat nearly the same list (single-cell, deconvolution, clinical use); keep one section and the canned refusal message.
Trim token overhead by cutting the "Implementation Checklist" (development bookkeeping) and condensing "When to Use", which restates the frontmatter description; this lifts the body toward the leaner anchor-4 conciseness profile.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly dense, task-specific reference material (argument, output, and error tables), but there is noticeable tightening potential: the out-of-scope list appears nearly verbatim twice ("When Not to Use" and "Do not use this workflow for" under Input Validation), "When to Use" restates the frontmatter description, and the closing "Implementation Checklist" is development bookkeeping rather than execution guidance. This sits between anchor 2 (several padded sections) and anchor 4 (minor trims only), so 3 fits best. | 3 / 5 |
Actionability | Guidance is largely copy-paste ready: full CLI invocations with all flags ("Rscript scripts/main.R --input_file ... --seed 42"), a complete argument table, concrete input/output examples, and per-error solutions. It misses anchor 5 because the Testing section and the bundled demo data point at files that are not present in the bundle ("tests/data/expression_matrix.csv", "tests/run_tests.R", "tests/test_skill.R"), so those commands are not executable as written. | 4 / 5 |
Workflow Clarity | The Workflow section gives a clear validate → run ESTIMATE → export sequence with an explicit validation step (numeric columns, unique sample names) and documented failure-preserving behavior for grouped comparison, matching anchor 4's clear sequence with most checkpoints. It falls short of anchor 5 because post-run verification is relegated to an "Optional post-check" that references a missing test script, so the verify→fix→retry loop is not fully closed. | 4 / 5 |
Progressive Disclosure | Structure is good: a "When to Read External Files" table maps situations to the real bundled files (references/algorithm.md, references/troubleshooting.md, references/cli-guide.md, scripts/main.R — all present) and references are one level deep, matching anchor 4. It is not a 5 because the body repeatedly points to a tests/ directory ("Need sample input files → tests/data/", plus the Testing section) that does not exist in the bundle, so navigation against the actual file listing has real gaps. | 4 / 5 |
Total | 15 / 20 Passed |