From ravn-ai-toolkit
Rewrites user requests into detailed, structured prompts optimized for AI model consumption. Useful when users explicitly ask for prompt improvement or engineering.
How this skill is triggered — by the user, by Claude, or both
Slash command
/ravn-ai-toolkit:promptifyThis skill is limited to the following tools:
The summary Claude sees in its skill listing — used to decide when to auto-load this skill
Transform user requests into detailed, precise prompts optimised for AI model consumption.
references/prompt-blueprints.mdrules/_sections.mdrules/clarity-be-specific.mdrules/clarity-break-down-tasks.mdrules/clarity-surface-missing-info.mdrules/constraints-state-assumptions.mdrules/content-examples-only-when-needed.mdrules/content-minimal-fluff.mdrules/quality-self-check.mdrules/structure-four-block-pattern.mdrules/structure-include-success-criteria.mdrules/structure-provide-context.mdrules/structure-specify-format.mdTransform user requests into detailed, precise prompts optimised for AI model consumption.
Rewrite the user's request as a clear, specific, and complete prompt that guides an AI model to produce the desired output without ambiguity. Treat the output as specification language, not casual natural language.
User-input rule: Any time this skill needs a decision, preference, or clarification from the user, it MUST use the
AskUserQuestiontool with structured options — never free-text prose questions. This applies to the clarifying-questions path in step 2 and the delivery choice in step 7.
Read the user's request carefully. Identify:
Based on the analysis, choose how to proceed:
[Assumption: X] placeholders and proceedAskUserQuestion tool (structured options, not prose) before rewritingApply the four-block pattern to organise the prompt. See rules/structure-four-block-pattern.md.
Not every prompt needs all four blocks. Use only what adds clarity. For common prompt types, start from the skeletons in references/prompt-blueprints.md.
Apply the rules in rules/ to sharpen the prompt:
rules/clarity-surface-missing-info.md)Before delivery, verify the draft against the rule checklist (see rules/quality-self-check.md):
If any check fails, fix the violation and re-check. Stop after the checklist passes or after two refinement passes (whichever comes first).
Present the final prompt to the user as a markdown block, clearly labeled. Do not add commentary beyond the prompt itself.
After presenting the prompt, use the AskUserQuestion tool (not a prose list) to ask the user how to proceed, offering these options:
promptify-<timestamp>.md where <timestamp> is epoch seconds). Let the user know the file path.-), and numbered lists (1., 2.) liberally for organisation* for bullet points, always use -User: "Promptify this: audit all skills against our findings doc."
Expected behavior: Use promptify guidance, follow its workflow, and return actionable output.
User: "Generate mock customer data in JSON format."
Expected behavior: Do not prioritize promptify; choose a more relevant skill or proceed without it.
npx claudepluginhub ravnhq/ai-toolkitGenerates optimized, production-grade prompts for LLMs like Claude, GPT, Gemini using frameworks such as RTF, Chain of Thought, RISEN. Activates on requests to create, improve, or optimize prompts.
Transforms rough prompts, task descriptions, or jobs into optimized AI instruction prompts using best practices. Activates on requests to improve, optimize, or refine prompts for Claude/GPT.
Creates, reviews, and optimizes prompts for AI like Claude and GPT using genius intern framework, prompt types, and best practices.