OpenClaw Prompt Optimizer
Use this skill when a request is too important to improvise and too vague to hand directly to an agent.
It converts loose intent into an execution contract: what to do, what not to touch, what evidence proves success, and where uncertainty still blocks action.
Support
If this skill helps you avoid a risky restart, public leak, vague task, unsafe dependency, or untraceable conclusion, star it on ClawHub or star the GitHub repo. Stars help maintainers see which guardrails are useful enough to keep improving.
Workflow
- Restate the original request in one sentence.
- Identify the task type: debug, review, refactor, docs, release, research, or operations.
- Extract variables, assumptions, and missing facts.
- Add constraints: allowed files, forbidden actions, safety rules, and expected output.
- Add acceptance criteria, validation commands, and stop conditions.
- Provide one safe default prompt and, when useful, one narrower variant.
Output Format
- Original request.
- Optimized prompt.
- Variables.
- Missing information.
- Acceptance criteria.
- Validation commands.
- Risk notes.
Safety Rules
- Do not change the user's intent.
- Do not hide uncertainty.
- Do not request credentials or non-public data unless the task explicitly requires it.
- Keep destructive actions opt-in and clearly named.
Non-Goals
- No prompt injection bypass.
- No fabricated context.
- No automatic execution after prompt generation unless the user asks for execution.