Contributing
Contributing
MAGIC Linguistic Agent Skills is open source under the Apache License 2.0. We welcome contributions — new skills, reference improvements, eval coverage, and documentation.
Quick Start
- Fork the repo on GitHub
- Clone and set up —
Python 3.10+,pip install -r requirements.txt - Make your changes — follow the skill anatomy and OpenSpec workflow
- Run tests —
pytest tests/(226 tests must pass) - Open a PR — include skill-judge scores and acceptance evidence
Key Resources
- CONTRIBUTING.md — full developer handbook: setup, pre-push checks, OpenSpec workflow, skill authoring tutorial, anti-patterns
- Skill Anatomy — universal SKILL.md structure
- Quality Gating — skill-judge 8-dimension 120-point rubric
- Evaluation Framework — eval authoring and scoring
Quality Gates
All contributed skills must meet skill-judge thresholds:
- Standard skills: ≥ B+ (96/120)
- Entry-point skills (orchestrator, scope, ethics, eval): ≥ A- (102/120)
- No P0 issues from skill-judge
Skill Patterns
Each skill follows one of these patterns:
| Pattern | When to use |
|---|---|
| Process | Step-by-step workflow with clear inputs/outputs |
| Navigation | Routing and coordination between skills |
| Mindset | Conceptual framing without prescribed steps |
| Tool | Wraps a specific tool or library |
| Philosophy | Domain principles and decision frameworks |
Security Vulnerabilities
If you discover a security vulnerability, do not open a public issue. Report it through GitHub Security Advisories.
Was this page helpful?
Edit on GitHub
Last updated on