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 detailed, actionable comparative-genomics skill with a clear six-phase workflow, concrete tool signatures, real fallback strategies, and a genuine bundled script. Weakest on conciseness, where a conceptual reasoning framework restates evolutionary-genomics knowledge Claude already has.
Suggestions
Trim the "Conservation Reasoning Framework" to tool-specific guidance and thresholds, dropping the general explanations of ortholog relationship types and dN/dS interpretation that Claude already knows — keep only the data path (`ensembl_get_homology` → CDS → `Sequence_dn_ds`) and the PhastCons/GERP/dN-dS cutoffs.
Make per-phase validation checkpoints explicit (e.g., "After Phase 2: confirm at least one 1:1 ortholog exists before proceeding to sequence retrieval") rather than leaving them implicit in the Fallback Strategies section.
Promote the full tool-parameter reference into a separate `references/tools.md` so SKILL.md stays a lean overview, with the key call signatures linked one level deep.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly efficient and tool-specific (exact parameters, return shapes, thresholds, fallbacks), but the "Conservation Reasoning Framework" section spends prose explaining concepts Claude already knows (1:1 vs 1:many orthologs, dN/dS interpretation, purifying selection) that could be tightened without losing the tool-specific value. | 3 / 5 |
Actionability | Provides concrete executable guidance throughout — `EnsemblCompara_get_orthologues(gene, species, target_species)`, `Sequence_dn_ds(seq1="ATG...", seq2="ATG...")`, `UniProt_search("gene:TP53 AND organism_id:9606 AND reviewed:true")`, and the bundled `scripts/dnds.py` with usage — with only minor gaps where calls are described at parameter level rather than as full runnable blocks. | 4 / 5 |
Workflow Clarity | A clearly sequenced six-phase workflow (with an ASCII pipeline diagram), per-phase reasoning, a Fallback Strategies section for error recovery, and Synthesis Questions acting as a review checklist; it is a read-only analysis workflow so the destructive/batch validation cap does not apply, but explicit per-phase validation checkpoints are mostly implicit rather than called out. | 4 / 5 |
Progressive Disclosure | Well-organized with clear section headers and one-level-deep, clearly signaled references (the bundled `scripts/dnds.py` and a sibling `tooluniverse-microbial-genome-characterization` skill) with no nested reference chains; the bulk of tool/parameter detail is inline, which is reasonable at this size but leaves minor organization gaps. | 4 / 5 |
Total | 15 / 20 Passed |