What problem does it solve?
Turning a recurring problem into a reusable Agent Skill often fails because requirements stay vague, the generated Skill is never tested against real behavior, and publishing to GitHub happens without authorization or verification. This Skill guides the full lifecycle from problem analysis to a validated, installable Skill directory.
Core Features & Use Cases
- Problem and behavior contracts: Converts vague ideas into explicit problem contracts, observable behavior contracts, and completion evidence before any files are generated.
- Skill scaffolding and validation: Generates a minimal Skill project with init_skill_project.py and checks frontmatter, links, placeholders, secrets, and script syntax with validate_skill_project.py.
- Graded evaluation: Defines four verification levels from structural checks to blind holdout/regression testing and verified npx installation, so quality claims match actual evidence.
- Use Case: You notice you repeatedly ask your Agent to review short-video hooks. Use this Skill to define the recurring problem, generate a hook-review Skill, test it against boundary and holdout samples, and optionally publish it to GitHub with a verified npx skills add command.
Quick Start
Ask the Agent to turn a problem you keep running into into an installable Skill, describing the recurring situation, the expected output, and what evidence would prove it works.
