prompt-preflight
Skill0 starsChanged 3 months agoApache-2.0View on GitHub
Silently refines ambiguous or underspecified user requests into the smallest actionable instruction that preserves intent, tone, scope, and creative wording.
In this repository46 other files it gives its agents.
- AGENTS.mdAGENTS.md
- academixSkill
- agent-browserSkill
- ai-engineering-platformSkill
- apocalypseSkill
- aria-apg-reviewSkill
- context-budgetSkill
- crawler-readiness-auditSkill
- data-protection-gateSkill
- decision-criticality-gateSkill
- fact-checkerSkill
- find-skillsSkill
- focus-friendly-outputSkill
- frontend-designSkill
- grill-meSkill
- grill-with-docsSkill
- humanizerSkill
- improve-codebase-architectureSkill
- microsoft-foundrySkill
- output-templatesSkill
- pre-launch-security-gateSkill
- quick-recapSkill
- rag-retrieval-qualitySkill
- rest-api-best-practicesSkill
- rigorous-responseSkill
- security-checklistSkill
- shepherdSkill
- skill-optimizerSkill
- sql-query-correctnessSkill
- system-design-scaleSkill
- tddSkill
- ui-ux-color-contrastSkill
- ui-ux-component-qualitySkill
- ui-ux-design-systemsSkill
- ui-ux-dropdownsSkill
- ui-ux-form-validationSkill
- ui-ux-generationSkill
- ui-ux-loading-statesSkill
- ui-ux-mobile-cssSkill
- ui-ux-typographySkill
- ux-logic-loopSkill
- ux-pattern-reviewSkill
- vercel-react-best-practicesSkill
- visual-flow-storyboardSkill
- visual-pr-reviewSkill
- web-design-guidelinesSkill
The file
We split this skill into the rules it gives an agent
7 rules. About 0 are common, 7 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈7
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 52
Do not add requirements the user did not imply.
- Distinctive: line 54
Use relevant context already available in the conversation or codebase.
- Distinctive: line 55
Keep tone, format, language, scope, and creative intent intact.
- Distinctive: line 56
Do not show the rewritten prompt unless asked.
- Distinctive: line 57
Do not delay execution just to rewrite.
- Distinctive: line 73
Verify against the original request, not just the rewrite.
- Distinctive: line 78
Keep refinement silent; return the preserved request or the single material clarification needed.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Prompt Preflight
Purpose
Silently decide whether a user request needs refinement before execution.
Refine only when the request is ambiguous, underspecified, contradictory, unstructured, or missing important execution constraints.
Operating Rules
- Preserve the user's intent.
- Do not add requirements the user did not imply.
- Fix obvious structure, wording, and sequencing issues.
- Use relevant context already available in the conversation or codebase.
- Keep tone, format, language, scope, and creative intent intact.
- Do not show the rewritten prompt unless asked.
- Do not delay execution just to rewrite.
- Execute the improved instruction immediately.
Skip Refinement When
- The request is already precise and actionable.
- The user explicitly wants literal execution.
- Rewriting would change creative intent or subjective wording.
- The task is a simple fact or direct command.
Preflight Loop
- Identify the intended outcome.
- Detect missing constraints or ambiguity.
- Build the smallest better instruction that preserves intent.
- Execute it.
- Verify against the original request, not just the rewrite.
Shared output profile
Compose the artifact profile from output-templates.
Keep refinement silent; return the preserved request or the single material clarification needed.
Preserve stricter domain rules above, omit empty components, and never fill a component with invented evidence.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

