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.
The body is well-structured, lean, and largely actionable, with ETE-specific details that genuinely add value. Its critical flaw is progressive disclosure: five bundle files exist but are completely unreferenced, so the detailed inlined material both duplicates them and leaves them undiscoverable.
Suggestions
Add a short 'References & Scripts' section linking references/workflows.md, references/api_reference.md, references/visualization.md, and the scripts/tree_operations.py and scripts/quick_visualize.py CLIs, so the bundle is discoverable from SKILL.md.
Move the detailed Implementation Details (format-code table, NHX serialization, event-detection steps, RF output explanation) into references/workflows.md and references/api_reference.md, keeping only one-line pointers in SKILL.md to remove the duplication.
In Example Usage, note that scripts/tree_operations.py already provides format conversion, rooting, pruning, and stats as a CLI, instead of re-implementing those steps inline; and add executable code (with method names) for the ortholog/paralog event-detection workflow, which is currently abstract steps.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly efficient: ETE-specific knowledge (format codes, NHX feature serialization, preserve_branch_length guidance) that Claude would not know, delivered as tight bullets and runnable code. It misses the 5 anchor because "When to Use" and "Key Features" restate the same capability list in two different forms, and the Implementation Details sections duplicate material that already exists in references/, so some tightening is possible. | 4 / 5 |
Actionability | Mostly executable guidance: a complete runnable example with pip install, format-code table, concrete write/features call, and specific method names (get_midpoint_outgroup, prune(preserve_branch_length=True), robinson_foulds). It stops short of the 5 anchor because the evolutionary event detection section gives only abstract numbered steps ("Provide a species naming function", "Run descendant event detection") with no method names or code, unlike the fully copy-paste-ready PDF extraction example. | 4 / 5 |
Workflow Clarity | The Example Usage is a clearly numbered 1–6 sequence with inline output checks (printing leaf/node counts), and event detection has a 4-step sequence. It sits at 4 rather than 5 because checkpoints are implicit (no verify-render-succeeded or error-recovery step) and the event-detection steps lack the concrete commands that would make them fully checkable; it is above the 3 anchor since sequences are explicit and operations are low-risk analysis/rendering, not destructive batch operations. | 4 / 5 |
Progressive Disclosure | The bundle contains references/api_reference.md, references/visualization.md, references/workflows.md, scripts/tree_operations.py, and scripts/quick_visualize.py, yet the body never mentions or links to any of them — no referenced paths exist anywhere in SKILL.md. Meanwhile the inlined Implementation Details (format codes, event detection, RF comparison) and the hand-rolled example duplicate content that clearly belongs in those separate files (workflows.md covers the same operations; tree_operations.py is a ready-made CLI for exactly the conversion/rooting/pruning steps shown), matching the 2 anchor ('content that clearly belongs in separate files is inlined; or references are buried'). It is not a 1 because the body itself is well sectioned and navigable. | 2 / 5 |
Total | 14 / 20 Passed |