Content
66%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 core content is a well-sequenced, highly actionable interactive ideation process with strong workflow gating and a useful verification checklist. Its two real weaknesses are meaningful redundancy across the overview, anti-patterns, and red-flags sections, and — more seriously — three referenced bundle files that do not exist, breaking the skill's progressive-disclosure structure.
Suggestions
Create the referenced files (frameworks.md, refinement-criteria.md, examples.md) in a references/ directory, or remove the instructions pointing to them — as written, Claude will be told to read files that don't exist.
Merge the 'Anti-patterns to Avoid' and 'Red Flags' sections (they list the same failure modes almost verbatim) and cut the 'How It Works' summary that restates the Process phases, to reclaim token budget.
Trim the Philosophy section's aphorisms (e.g. "Show people the future — don't just give them better horses") down to the operative guidance, and drop the 'Trigger Phrases' block that duplicates the frontmatter description.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Mostly useful guidance, but noticeably redundant: the "How It Works" section restates the Process phases, the "Anti-patterns to Avoid" and "Red Flags" sections repeat the same failure modes almost verbatim (e.g. "Generating 20+ shallow variations", "yes-machining", "Skipping the 'who is this for' question"), and Trigger Phrases duplicate the frontmatter. This fits the 'could be tightened' 3 anchor rather than the 4 anchor, where only minor trimming would be needed. | 3 / 5 |
Actionability | Concrete and executable for an instruction-only skill: specific lens questions, a 3-5 question cap, named tools (AskUserQuestion, Glob, Grep, Read), an exact output path (`docs/ideas/[idea-name].md`), a copy-paste-ready one-pager template, and an init script. Not a 5 because three referenced guidance files (frameworks.md, refinement-criteria.md, examples.md) do not exist in the bundle, leaving those instructions unexecutable, and the script path assumes a fixed install location. | 4 / 5 |
Workflow Clarity | The three phases are clearly sequenced with explicit checkpoints and feedback loops: "Do NOT proceed until you understand who this is for and what success looks like", the phase transition gated on user reaction, "Only save if they confirm", and a closing Verification checklist. This is a direct match for the 5 anchor; nothing is missing for a conversational skill with no destructive or batch operations. | 5 / 5 |
Progressive Disclosure | The body tells Claude to "Read `frameworks.md`", "Read `refinement-criteria.md`", and "Read `examples.md`" in this skill directory, but the bundle contains only `scripts/idea-refine.sh` — no references/ directory and none of the three files exist, so every external reference is dangling. Navigation is therefore broken despite clear in-body structure, which lands at the 2 anchor rather than 3 (where references would at least exist but be poorly signaled). | 2 / 5 |
Total | 14 / 20 Passed |