// Forces two things on every non-trivial request: restate the user's intent before answering, and take a super-dimensional view — name the domain, search for existing methods and prior art, bring specialist knowledge in before improvising.
| name | promptly-prompt |
| description | Forces two things on every non-trivial request: restate the user's intent before answering, and take a super-dimensional view — name the domain, search for existing methods and prior art, bring specialist knowledge in before improvising. |
AI answer quality depends on context quality, not prompt tricks.
This skill operates via a UserPromptSubmit hook that injects two
disciplines into complex requests:
Restate, then pause if needed — echo the user's request in your own words, surface implicit constraints, name what would make the answer wrong. If anything is ambiguous or your planned approach might not be endorsed, stop and ask before acting. If the restatement makes it clear there is no disagreement, continue. Skip the restatement only when the request is genuinely trivial.
Super-dimensional view — before solving from memory, name the domain, search for established methodologies, frameworks, libraries, and prior art, then bring that specialist knowledge to bear. Cite specific names, not vague gestures. If nothing fits, say so and explain why.
The hook script at scripts/intercept.py scores prompt complexity using
rule-based signals. Simple commands pass through untouched. Complex requests
get the full injection.
When invoked directly (e.g., user says "optimize this prompt"), apply the two disciplines manually: restate the user's intent with implicit needs surfaced, then locate the domain and the existing methods that belong to it. Rewrite the prompt with that context filled in.