1ai-skills

Your AI agent is lazy. These 1337 skills fix that.

AI agents skip tests, ignore security, write placeholder code, and cut corners whenever they can. 1ai-skills forces them to follow real workflows with anti-rationalization tables, code examples, and verification checklists — across every domain.

The Problem

Your AI agent says:

"I'll add tests later." "Security is a nice-to-have." "This placeholder works fine."

It doesn't. It won't. It never does.

The Solution

Every skill in this library includes an Anti-Rationalization Table — a structured argument that prevents your agent from cutting corners:

| Rationalization | Reality |
|---|---|
| "Tests slow me down" | Bugs slow you down 10x more. Tests are speed, not overhead. |
| "I will refactor later" | Technical debt compounds. Refactor as you go. |
| "It works on my machine" | If it is not in CI, it does not work. Ship proof, not claims. |

Quick Start

Claude Code (Recommended)

/plugin marketplace add oyi77/1ai-skills
/plugin install 1ai-skills@1ai-skills

Cursor

git clone https://github.com/oyi77/1ai-skills.git
cp -r 1ai-skills/cybersecurity/* .cursor/rules/

Gemini CLI

gemini skills install https://github.com/oyi77/1ai-skills.git --path skills

Any Agent

git clone https://github.com/oyi77/1ai-skills.git
# Point your agent to the skills/ directory

1337 Skills Across 19 Categories

What Makes This Different

Every Skill Has

• YAML frontmatter — Machine-readable metadata (name, description, domain, tags)
• Anti-Rationalization Table — Prevents agents from cutting corners
• Code Examples — Concrete, runnable code (Python, JS, Bash, SQL)
• Workflow — Step-by-step instructions with bold labels
• When to Use — 3+ trigger conditions
• When NOT to Use — 2+ contraindications
• Verification Checklist — Success criteria you can check

Test Suite

python3 scripts/test-skills.py           # Full test
python3 scripts/test-skills.py --quick   # Skip SDK checks
python3 scripts/test-skills.py --json    # JSON output
python3 scripts/test-skills.py --skill NAME  # Single skill

8 test dimensions:

1. Structure (YAML frontmatter, required fields)
2. Content (sections, depth, no placeholders)
3. Code syntax (Python ast.parse, JS/TS, Bash)
4. Internal links (all /skills/ links resolve)
5. Description quality (50-200 chars, action-oriented)
6. Quality markers (anti-rationalization, code, verification)
7. SDK availability (referenced imports are installable)
8. Workflow completeness (has workflow section)

Result: 1337/1337 PASS · 0 warnings · 0 failures

New Skills (Just Added)

Self-Evolving System

1ai-skills includes 13 meta-skills that make the library improve itself:

• skill-tracker — Logs every skill invocation with metrics
• skill-evolver — Analyzes usage patterns and suggests improvements
• skill-feedback-capture — Captures user feedback on skill quality
• auto-learner — Learns from corrections and updates skills
• performance-monitor — Tracks skill success rates and response quality
• skill-banner — Shows activation banner with skill summary

Slash Commands

For Contributors

# Create a new skill
python3 scripts/generate-site.py          # Regenerate website
bash scripts/audit-skills.sh --write      # Update SKILLS.json
python3 scripts/validate-skills.py        # Validate structure
python3 scripts/test-skills.py            # Run all tests

See CONTRIBUTING.md for guidelines.

Compatible With

• Claude Code
• Cursor
• Gemini CLI / Antigravity
• Windsurf
• OpenCode
• GitHub Copilot
• Kiro
• Any agent that reads SKILL.md files

License

MIT — use freely, contribute back.

Star this repo if it saves your agent from being lazy.