Content
75%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 highly actionable body: executable verified tool calls with real IDs, concrete source-selection guidance, and well-sequenced reasoning frameworks with useful checkpoints. The main weakness is token efficiency — the community/evolutionary ecology sections restate textbook concepts Claude already knows instead of adding tool-specific or non-obvious guidance.
Suggestions
Trim the definitional lines in sections 5 and 6 (Gause's principle, aposematism, Batesian vs Mullerian, r/K selection, trophic cascades) down to tool-oriented guidance — e.g., which tool answers a keystone-species or mimicry question — since Claude already knows these concepts.
Merge the pollinator-ecology statistical notes with the population-dynamics survival-analysis guidance into one 'statistical methods' section to avoid repeating analysis advice across sections.
Move the 8-source additional-taxonomy table and its four worked examples into a reference file, keeping only the source-selection decision guide inline in SKILL.md.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The tool tables, ID-interchangeability warnings, and verified examples are dense and non-obvious, but sections 5–6 restate textbook ecology Claude already knows ("Competitive exclusion: Two species cannot stably coexist on single limiting resource (Gause's principle)", "Aposematism: Warning coloration signals toxicity/unpalatability", "Batesian (harmless mimics dangerous) vs Mullerian"). This fits the 3 anchor 'mostly efficient but includes some unnecessary explanation' — not 2, because most content (API gotchas, source-selection rules) earns its tokens. | 3 / 5 |
Actionability | Fully executable, copy-paste-ready calls with real verified IDs ("tu.run_tool(\"GBIF_get_taxon_parents\", {\"taxon_key\": key})" with usageKey 5219404, tsn 183803, MRGID 25180) plus concrete search templates ("[species] invasive impact [region]") and per-source selection rules covering common cases. This matches the 5 anchor — the examples even document parameter-shape gotchas like OpenTree's comma-separated `names` string. | 5 / 5 |
Workflow Clarity | Multi-step processes are clearly sequenced ("Resolve a name to a GBIF usageKey once, then navigate the Backbone tree") with useful checkpoints such as "check the returned `rank` before assuming species-level". Matches the 4 anchor 'clear sequence with most checkpoints present' — not 5 because there are no explicit validate-and-retry loops, though no destructive or batch operations require them. | 4 / 5 |
Progressive Disclosure | A single-file skill with no bundle files, well-organized into clearly headed sections, a decision guide for choosing among overlapping identity sources, and no nested or buried references. The two large tool tables could arguably live in reference files, which is the minor organization gap of the 4 anchor rather than the 'should be separate is inline' situation of 3. | 4 / 5 |
Total | 16 / 20 Passed |