Skills
Skill execution model and where skill definitions live in Synap backend/OpenClaw flows.
Skills package reusable agent capabilities into auditable execution units.
What a skill does
- defines bounded actions and expected behavior
- maps calls to Hub Protocol/API operations
- keeps high-impact writes reviewable through governance
Where skills are stored
Backend repository locations:
- Synap OpenClaw skill spec:
skills/synap/SKILL.md - Companion docs:
skills/synap/README.md
These are the concrete documents to reference when aligning agent behavior with current backend contracts.
Invocation path
At runtime, skills typically call into Hub endpoints:
/api/hub/*for structured operations/v1/chat/completionswhere OpenAI-compatible streaming is needed
Primary backend contracts:
- Hub router:
packages/api/src/routers/hub-protocol-rest.ts - Chat-completions bridge route:
packages/api/src/routers/channels.ts