writing-skills

Create and test agent skills using test-driven development with subagent pressure scenarios.

Updated Aug 22, 2026
One-click install
npx skills add https://github.com/farslab/claude-skills --skill writing-skills-farslab
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: writing-skills
Source: https://github.com/farslab/claude-skills/tree/main/writing-skills
Command: npx skills add https://github.com/farslab/claude-skills --skill writing-skills-farslab

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires graphviz.

What problem does it solve? Writing documentation-style skills for AI agents often produces instructions that agents ignore or rationalize away under pressure. This Skill applies test-driven development to skill authoring so you can verify agents actually comply before deploying. ## Core Features & Use Cases - TDD-based skill authoring: Run baseline pressure scenarios with subagents (RED), write the skill addressing observed failures (GREEN), then close rationalization loopholes (REFACTOR). - Skill Discovery Optimization (SDO): Guidance for writing description fields, keyword coverage, naming conventions, and token-efficient content so agents find and load the right skill. - Testing methodology: Detailed reference on pressure scenario design, rationalization tables, red flags lists, and meta-testing techniques. - Use Case: You want to create a new discipline-enforcing skill (e.g., requiring verification before completion). You run baseline scenarios without the skill, document the agent's exact rationalizations, write the skill to counter them, and re-test until the agent complies under maximum pressure. ## Quick Start Ask the agent to help you create a new skill by first running a baseline pressure scenario with a subagent, then drafting the SKILL.md based on the observed failures.

Frequently Asked Questions about writing-skills

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I write a SKILL.md file for an AI agent?▼

A SKILL.md requires YAML frontmatter with name and description fields, followed by a markdown body with an overview, when-to-use triggers, core patterns, and examples. Keep the description focused on triggering conditions starting with "Use when" rather than summarizing the workflow.

How do I test that an agent skill actually works?▼

Test skills by running pressure scenarios with subagents: first without the skill to document baseline failures, then with the skill to verify compliance. Combine multiple pressures like time constraints, sunk cost, and authority to surface rationalizations the skill must counter.

What makes a good skill description for agent discovery?▼

A good description starts with "Use when", describes only triggering conditions and symptoms in third person, and never summarizes the skill's workflow. Descriptions that summarize processes cause agents to follow the description instead of reading the full skill.

When should I not create a new skill?▼

Skip creating skills for one-off solutions, project-specific conventions, standard practices documented elsewhere, or mechanical constraints that can be enforced with validation scripts. Skills are for reusable techniques, patterns, and references that apply broadly.

Why do agents ignore or bypass skill instructions?▼

Agents rationalize around rules under pressure using excuses like "spirit versus letter" or "being pragmatic." Counter this with explicit loophole closures, rationalization tables, red flags lists, and foundational principles stated early in the skill.