Content
92%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-engineered instruction-only skill: a precise five-step workflow with real validation checkpoints (including a runnable self-check script), a complete worked example, and a clean split between the overview and a single one-level-deep reference file whose contents match the body's promises. The only weakness is mild: a few sentences of rationale in the body duplicate material already in the reference file.
Suggestions
Trim the 'The philosophy' section to its operative rules (observable-done corollary, brand-profile note) and move the persuasive rationale for the seven-part anatomy entirely into references/goal_prompt_patterns.md, which already covers it under 'Why each part exists'.
Shorten or drop the trailing derivation footer ('Derived from duncan-buildroom/freeskills...') — provenance adds tokens without changing execution.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense and directive, and it correctly pushes the 'why' rationale (per-part explanations, failure-mode detail) into references/goal_prompt_patterns.md. Minor over-explanation remains: the 'The philosophy' section's rationale ('Get out of the model's way', the 'Make it good is a wish' illustration) partially duplicates the reference file's 'Why each part exists', and the derivation footer could be trimmed — so it sits above 'mostly efficient' (3) but short of 'every token earns its place' (5). | 4 / 5 |
Actionability | Guidance is fully concrete and executable: six named extraction slots, a one-batch question rule, a verify step with literal actions ('ls the paths, glance at the available-tools list'), a quoted verbatim goal-line template, a 150-350 word target, a runnable check ('python3 scripts/goal_prompt_self_check.py -' on stdin, exit 0 semantics), and a complete worked example covering the common case. Nothing is pseudocode or hand-wavy; the instruction-only nature carries no penalty since the guidance is directly actionable. | 5 / 5 |
Workflow Clarity | A clearly sequenced five-step process (extract → fill gaps/ask once → verify resources → write → self-check and deliver) with explicit validation checkpoints: the step-3 resource verification, the six-point self-check checklist, and a mechanical script check with a fix-any-miss-before-shipping feedback loop. This matches the anchor for clear sequence + explicit validation + error-recovery loop + checklist; the destructive/batch cap does not apply since the skill produces text, not risky operations. | 5 / 5 |
Progressive Disclosure | SKILL.md is a 79-line overview; per-medium verification defaults, the failure-mode catalog, and the second worked example live one level deep in references/goal_prompt_patterns.md (verified to exist and contain exactly what the links promise), and the mechanical self-check lives in scripts/goal_prompt_self_check.py (verified to match its described behavior). Links are inline, clearly signaled, and state what each file contains; no inlined content belongs in a separate file. | 5 / 5 |
Total | 19 / 20 Passed |