Claude skill prompt-master trends on GitHub: generates precise prompts for any AI tool

The repo is distributed as a Claude skill, installable via claude.ai's Customize → Skills by uploading a ZIP, or by cloning into ~/.claude/skills. Each request runs a fixed pipeline: identify the target tool, extract intent across 9 dimensions (task, input, output, constraints, context, audience, memory, success criteria, examples), ask up to 3 clarifying questions when key info is missing, then auto-match one of 13 prompt templates (RTF, CO-STAR, RISEN, CRISPE, etc.) without exposing the template name, and finally run a token-efficiency pass to trim words that don't affect output. It ships 20+ tool profiles covering Claude Code, Cursor, GitHub Copilot, Midjourney, Stable Diffusion, ComfyUI, Sora, ElevenLabs, and Zapier, with a 4-question generic fingerprint for anything else. For developers writing prompts all day, it saves the repeated API calls spent rewriting; for those editing code in Cursor or Windsurf, it bakes file paths, function names, and do-not-touch lists into the prompt to narrow unintended edits; for Midjourney and Stable Diffusion users, examples lock in 16:9 aspect ratio, --v 6, and negative prompts to reduce style drift. Note that it claims to verify 'latest' model parameters against official docs, meaning the skill itself must be updated as models iterate, or the routed framework may not match the current version's interface behavior.