Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →notque/multi-persona-critique
Critique a written proposal or design artifact via 5 philosophical personas in parallel, with consensus synthesis.
Passed 7 of 18 test cases with the skill, and 2 of 18 without it.
and 9 more in the full report
6 fixed2 passed either way13 still failing1 broke
Tokens, turns and time describe resource use, not a dollar price.
One of 6 test cases the skill fixed. Same model, same prompt, with and without the skill.
I need a comprehensive critique of three proposed database caching strategies for our web service: 1) Client-side IndexedDB caching, 2) Redis read-aside cache with write-through, and 3) Edge CDN stale-while-revalidate. Please evaluate all three through 5 separate intellectual lenses—such as logical soundness, practical implementation burden, systems design elegance, end-user experience, and long-term unintended consequences. Have each evaluator independently rate and rank every proposal, then synthesize their findings into a consensus report complete with an evaluation matrix, ranked recommendations, and an analysis of key disagreements between reviewers.
| Checked | Without skill | With skill |
|---|---|---|
| The evaluation assigns every proposal a rating from the allowed set STRONG, PROMISING, WEAK, or REJECT. | Missed | Passed |
| The weighted score mapping assigns STRONG as 3 points. | Missed | Passed |
| The report identifies Bertrand Russell as The Logician persona. | Missed | Passed |
The instructions your model reads when the skill loads.
This skill takes a set of proposals — feature ideas, architectural decisions, design choices, strategy options — and sends ALL of them to 5 distinct intellectual personas for parallel, independent critique. Each persona brings a different philosophical lens. The skill then synthesizes all critiques into a consensus report showing where personas agree, where they disagree, and what the disagreements reveal.
This is NOT the roast skill. Key differences:
roast critiques CODE with HackerNews personas and validates file:line claimsroast is evidence-based (checking actual code); this is argument-based (evaluating reasoning)Key constraints baked into the workflow:
| Signal | Load These Files | Why |
|---|---|---|
| example-driven tasks, errors | examples-and-errors.md | Loads detailed guidance from examples-and-errors.md. |
| dispatching the five critique personas (Phase 2) | personas.md | Loads detailed guidance from personas.md. |
| writing the Phase 5 synthesis report | synthesis-template.md | Loads detailed guidance from synthesis-template.md. |
| roast-style code critique with HN personas | roast.md | Loads the roast workflow: evidence-based critique via 5 HN personas with claim validation. |
| HN persona specs, claim format, validation verdicts | hn-personas.md | Loads HN commenter persona specifications, prompt template, and claim format. |
Goal: Extract or generate clear, numbered proposals ready for critique.
Step 1: Determine input mode
test cases passed, vs 2 / 18 without
More skills from the same source and nearby in the registry, with their headline benchmark lift.