What problem does it solve?
Creating effective AI skills requires more than just writing documentation - it demands testing under pressure, bulletproofing against agent rationalization, and optimizing for discovery. Without a systematic approach, skills remain untested, ambiguous, and ignored when they matter most.
Core Features & Use Cases
- TDD for Documentation: Applies test-driven development to skill creation with RED-GREEN-REFACTOR cycles, baseline testing with subagents, and pressure scenarios that reveal compliance failures.
- Bulletproofing Techniques: Provides explicit counters for agent rationalizations, red flag lists, and foundational principles that prevent discipline-enforcing skills from being bypassed under pressure.
- Discovery Optimization: Teaches how to write descriptions that trigger correctly, use keyword coverage for searchability, and structure skills for progressive disclosure to minimize token usage.
Quick Start
Use the writing-skills skill to create a new skill by first running a baseline pressure scenario without the skill, documenting the agent's failures, then writing minimal documentation that addresses those specific violations before re-testing.
