Delivered project
Prompt Engineering: a skill for AI agents
Open source skill that turns a raw draft into a production-grade prompt, picking techniques by task type and target model.
- Case
- prompt engineering skill
- Code
- Open source
- Stack
- Claude Code / Agent Skills / Prompt engineering

§ Overview
The skill diagnoses the draft, applies a canonical structure (role, context, instructions, examples, output contract and success criteria) and returns a ready-to-paste prompt with a short changelog of what changed and why. It runs in Claude Code, GitHub Copilot, Codex, Gemini CLI, OpenCode, Cursor, Windsurf and Cline, and ships through the Prompt Cause marketplace.
- Claude Code
- Agent Skills
- Prompt engineering
- Markdown
- GitHub Actions
B/01
Challenges
- Prompts written in a rush, with no structure or success criteria
- Each model responds best to different techniques
- A different install flow for every agent and IDE
B/02
Solutions
- Draft diagnosis before rewriting
- Technique selection by task type and model (Claude, OpenAI, Gemini)
- Own marketplace with CI validation and native install per agent
B/03
Features
- Rewrite with canonical structure and paste-ready output
- Changelog of what changed and why
- Calibration for reasoning and chat models
- MIT license and open contributions