Building consistent LLM outputs is harder than it looks, and this skill tackles it head-on. You get a structured approach to creating prompt templates with strict contracts, including system and user prompts, JSON schemas for validation, few-shot examples, and explicit do/don't lists. The templates support variable interpolation, versioning, and chaining multiple prompts together. What I appreciate is the focus on output contracts with Zod schemas, so you're not just hoping the LLM returns valid JSON. If you're tired of brittle prompts that work once then fail mysteriously, or you need to maintain prompt quality across a team, this gives you the scaffolding to treat prompts like real code with types and tests.
npx -y skills add patricio0312rev/skills --skill prompt-template-builder --agent claude-codeInstalls into .claude/skills of the current project.
Select a file.
sickn33/antigravity-awesome-skills
github/awesome-copilot
github/awesome-copilot