testing-skills-with-subagents

Test skill compliance using a RED-GREEN-REFACTOR TDD cycle.

10|Updated Dec 15, 2014
One-click install
npx skills add https://github.com/baleen37/dotfiles --skill testing-skills-with-subagents-baleen37
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: testing-skills-with-subagents
Source: https://github.com/baleen37/dotfiles/tree/main/modules/shared/config/claude/skills/testing-skills-with-subagents
Command: npx skills add https://github.com/baleen37/dotfiles --skill testing-skills-with-subagents-baleen37

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

Skills, especially those enforcing discipline, can be bypassed by agents under pressure or through rationalization. This skill applies a Test-Driven Development (TDD) cycle to process documentation, ensuring skills are robust, resist loopholes, and guarantee compliance even in high-stakes scenarios.

Core Features & Use Cases

  • TDD for Documentation: Uses a RED-GREEN-REFACTOR cycle to test skill effectiveness, starting by observing agent failures without the skill.
  • Pressure Scenario Testing: Creates realistic scenarios with combined pressures (time, sunk cost, authority) to stress-test skill compliance.
  • Rationalization Countering: Identifies and explicitly plugs loopholes by documenting agent excuses and adding specific negations to skill rules.
  • Use Case: Before deploying a new TDD skill, you'd run a scenario where an agent is tempted to skip tests. This skill helps you capture the agent's rationalizations and refine the TDD skill until the agent consistently follows the rules.

Quick Start

1. RED Phase: Baseline Testing

- Create pressure scenarios (3+ combined pressures).

- Run WITHOUT skill, document agent choices and rationalizations.

2. GREEN Phase: Write Minimal Skill

- Write skill addressing specific baseline failures.

- Run WITH skill, verify compliance.

3. REFACTOR Phase: Close Loopholes

- Capture new rationalizations.

- Add explicit negations, update rationalization table and red flags.

- Re-verify until bulletproof.