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, tool-rich reasoning skill with concrete data-source guidance and sequenced workflows anchored by a 'look up, don't guess' verification philosophy. Main weaknesses are restating textbook neuroscience Claude already knows and giving a Python directive with no code examples.
Suggestions
Trim or move well-known textbook material (Nernst/Goldman equations, action-potential phases, cranial-nerve list, UMN/LMN table) to a reference file or condense to the non-obvious specifics, keeping the context budget for tool usage and pitfalls.
Add at least one executable Python snippet under the computational neuroscience section, since the skill repeatedly instructs to 'ALWAYS use Python' but shows no code.
Consider extracting the tools table and the dense equation/anatomy reference into a references/ file, leaving SKILL.md as a concise overview that points to it for detail.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The bullet/equation format is lean rather than padded prose, but the body restates substantial textbook knowledge Claude already has (Nernst equation, action-potential phases, basal ganglia pathways, cranial-nerve mapping, UMN/LMN table) alongside genuinely novel content like tool names and the ASJ->PVQ connectome fact. | 3 / 5 |
Actionability | The tools table gives concrete invocations with parameters and the search-query templates ('[region A] projection [region B]') are executable, but the 'ALWAYS use Python' directive lacks any accompanying code example, a notable gap for a computation-emphasizing skill. | 4 / 5 |
Workflow Clarity | Numbered reasoning patterns (quantitative strategy, 'where in the brain', clinical neuro) include explicit 'VERIFY with PubMed search' checkpoints, but validation is framed as optional lookup rather than a hard checkpoint with error-recovery feedback loops. | 4 / 5 |
Progressive Disclosure | Well-organized into seven clearly-headed sections with no nested references, but as a single 185-line file with dense reference material (tools table, equation sets, anatomy tables) that could plausibly be split into reference files, it has minor organization gaps. | 4 / 5 |
Total | 15 / 20 Passed |