Content
88%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 content is highly actionable with executable, parameterized tool calls and a clearly phased workflow backed by a verification checklist. The main weakness is mild verbosity in the interpretive prose and a single-file structure that, while well-organized, could offload some reference material.
Suggestions
Tighten the Interpretation Table: drop definitional prose Claude already knows (e.g., what N50 is, typical bacterial genome size) and keep only the domain-specific interpretation tied to NCBI fields.
Consider moving the three Worked Examples into a references/ file (e.g., EXAMPLES.md) with a one-line pointer from SKILL.md to improve progressive disclosure and reduce inline length.
The 'COMPUTE, DON'T DESCRIBE' directive is good but generic; tie it to a concrete pandas snippet showing the Phase 5 sort-by-(level, N50, contig count) pattern so the guidance is immediately executable.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly efficient with concrete tool calls and minimal fluff, but the Interpretation Table and Worked Examples contain explanatory prose (e.g., defining Contig N50, what 'Complete Genome' means, typical bacterial genome size) that assumes less domain knowledge than necessary and could be trimmed. Not a 3 because it is not padded with generic concepts Claude already knows; not a 5 because some explanatory sentences could be cut. | 4 / 5 |
Actionability | Fully executable copy-paste-ready tool invocations with exact JSON parameters appear in every phase and in three worked examples with real accessions, covering the common cases — matching the top anchor. | 5 / 5 |
Workflow Clarity | Phases 0–5 are clearly sequenced with explicit selection heuristics (Phase 2), verification directives ('LOOK UP, DON'T GUESS', 'metadata.total_available', prefer sequence_reports over number_of_chromosomes), and a final Completeness Checklist as an explicit validation step. The operation is read-only so the destructive-cap caveat does not apply. | 5 / 5 |
Progressive Disclosure | The body is well-organized with clear section headers (Tools, Workflow, Interpretation, Worked Examples, Limitations, Checklist) and no nested references, and no bundle files exist so all content is intentionally inline. Scored 4 rather than 5 because the skill exceeds 50 lines and the long Interpretation Table and three Worked Examples could arguably be split into reference files, leaving minor organization headroom. | 4 / 5 |
Total | 18 / 20 Passed |