v0.2.0 — Agents should execute whenever possible

Registry v0.2.0

Agents should execute whenever possible.

Open registry of reusable AI agent skills and capability definitions.

Stats

• 141 capabilities across 15+ domains
• 36 ready-to-use skills
• Governance guardrails with automated overlap detection
• Capability sunset policy enforcement

What's New in 0.2.0

• 19 new capabilities added
• Domain reviews: text, web, model, agent
• Status labels (draft → stable)
• Vocabulary expansion
• YAML parsing fixes
• Community files: CONTRIBUTING, CODE_OF_CONDUCT, SECURITY
• Issue templates for bugs and features

Quick Start

\\�ash
pip install -r requirements.txt
python tools/validate_registry.py
python tools/generate_catalog.py
\\

Companion Runtime

See agent-skills for the execution engine.

License

Apache 2.0