Content
60%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 Workflow section is a genuinely actionable, well-sequenced pipeline with real script commands and explicit decision gates, but it is undermined by roughly a third of the body being auto-generated boilerplate (When to Use, Key Features, Implementation Details) that duplicates the frontmatter description and buries the useful content. Trimming to the Workflow, Dependencies, and a fixed Example Usage would make this a compact, high-quality skill.
Suggestions
Delete the "When to Use", "Key Features", and "Implementation Details" boilerplate sections — they restate the frontmatter description verbatim and add generic filler; keep only the Workflow, Dependencies, and Example Usage sections.
Replace the hardcoded dated path in Example Usage ("cd 20260316/scientific-skills/Data Analytics/meta-feasibility-analyzer") with a path-agnostic instruction, and remove the dangling "See `## Workflow` above for related details" self-reference.
Add brief error-recovery guidance to the Workflow (e.g., what to do if the PubMed search fails or the extract step returns no query), which would also lift workflow clarity toward the top anchor.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Three sections are boilerplate padding: "When to Use" (which restates the description with broken grammar, "when you need analyzes the feasibility"), "Key Features" ("Scope-focused workflow aligned to: Analyzes the feasibility..." duplicates the description verbatim), and "Implementation Details" (generic filler like "Execution model: validate the request, choose the packaged workflow" that adds no skill-specific information). This matches anchor 2's "noticeably verbose; several unnecessary explanations or padded sections" — the ~100-line Workflow core is lean, but a third of the body is duplicative template text, keeping it below anchor 3's "mostly efficient". | 2 / 5 |
Actionability | The workflow provides real, executable commands that match the actual script subcommands ("python scripts/feasibility_ops.py extract --text", "search --query ... --type clinical", "clinical --json ... --query", "meta --json") plus fully specified LLM prompts with rules and an example output. It is not a 5 because the commands are templates with {{placeholders}} rather than a concrete worked example with real values, and the Example Usage section relies on a hardcoded dated path ("cd 20260316/scientific-skills/...") that will not exist for most users. | 4 / 5 |
Workflow Clarity | The nine steps are clearly sequenced with an explicit decision gate at step 5 ("If clinical_count == 0: ... STOP") and explicit final decision logic at step 8 (no meta-analyses + trials → feasible; new trials after latest meta-analysis → update feasible; else not feasible). This is a read-only search workflow so the destructive/batch validation cap does not apply, but it stays at 4 rather than 5 because there is no error-recovery guidance for failure modes such as an API error or an empty extraction result. | 4 / 5 |
Progressive Disclosure | The single bundle file referenced (scripts/feasibility_ops.py) exists, matches the body's commands, and references are one level deep — appropriate for a self-contained skill of this size. It falls short of anchor 5's "easy navigation" due to organization warts: the dangling self-reference "See `## Workflow` above for related details" in Implementation Details, and the redundant boilerplate sections that bury the actual Workflow content below the fold. | 4 / 5 |
Total | 14 / 20 Passed |