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, genuinely actionable instruction skill: clear workflow, strong anti-fabrication rules, a real template and reference guide, and a concrete worked example. The main cost is token weight — five generic boilerplate sections at the end and a repetitive example inflate the body without adding protocol-specific guidance.
Suggestions
Cut or merge the generic 'Output Contract', 'Failure Handling', 'User Checkpoints', 'Input Validation', and 'Quick Validation' sections into a short, protocol-specific validation checklist, saving roughly 30 lines of boilerplate.
Trim the worked example to 2–3 parameter fields marked 'To be supplemented/Not provided' — the convention is established after a few uses — or move the full example into references/guide.md.
Make 'Validate logical sequence and operability' concrete by listing the specific checks (step ordering, unit consistency, missing-parameter flags) that Output Check should perform.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The protocol-specific content (When to Use, Implementation Details, the worked example) is efficient and earns its tokens, but the tail sections — 'Output Contract', 'Failure Handling', 'User Checkpoints', 'Input Validation', 'Quick Validation' — are generic scaffold boilerplate (e.g., 'Primary output: Structured result or target file aligned with this skill's objective') that adds little skill-specific value, and the example repeats 'To be supplemented/Not provided' fifteen-plus times where a few instances would establish the convention. This lands on anchor 3 (mostly efficient, some tightening needed) rather than 2 because the padding is concentrated in a few sections rather than permeating the document. | 3 / 5 |
Actionability | Concrete guidance throughout: a four-step workflow, explicit parameter rules ('Never fabricate values', unit conventions), real bundle paths (assets/protocol_template.md, references/guide.md), an output naming convention ('{Experiment_Info_Abbreviation}.txt' in outputs/), and a full worked input→output example. It is not a 5 because some directives stay abstract — 'Validate logical sequence and operability' gives no check procedure — leaving minor gaps per anchor 4. | 4 / 5 |
Workflow Clarity | The sequence is explicit and ordered (Step Review → Parameter Completion → Standardization → Output Check), with validation present via 'Output Check: Validate logical sequence and operability', the 'Critical Parameters to Supplement' list, and a Quick Validation section, plus user checkpoints before batch operations. It misses anchor 5 because the error path is stated generically ('report the failing step... recovery suggestions') rather than as an explicit validate→fix→retry feedback loop. | 4 / 5 |
Progressive Disclosure | The bundle structure is sound: SKILL.md is the overview, assets/protocol_template.md (a fill-in template) and references/guide.md (checklists) both exist and are clearly signaled by path, one level deep. It is not a 5 because the ~50-line worked example output inlined in SKILL.md is content that could be split into references, and guide.md partially duplicates body content rather than adding a distinct layer. | 4 / 5 |
Total | 15 / 20 Passed |