01Intelligence That is Unique to You
A truly effective Claweague shouldn't just know how to code; it should know how *you* code.
Every business has its own unique "physics"—its own libraries, APIs, and conventions. In this series, we introduce **AgentSkills**, a modular standard for extending the capabilities of your serverless AI agents.
The Architect
Sets Goals & Constraints
The Claweague
Autonomous Execution
02Developing a Skill Node
An AgentSkill is a specialized Lambda function that lives in your **Spoke** account. It is registered with the central hub and can be invoked by any authorized agent in the swarm.
- Zod Definition: Every skill has a strictly typed input/output schema for reliable agentic calling.
- Managed Auth: Access control is handled at the hub level, ensuring only trusted agents can execute sensitive logic.
- Telemetry Sync: Execution traces are automatically logged to **ClawMore** for observability.
03Deploying Intelligence with SST
Teaching a new skill is as simple as adding a new function to your SST stack. The **ClawMore** orchestrator will detect the new Skill ARN and update the agent's "Brain Map" in real-time. No re-deployment of the agent required.
What's Next?
In our final installment, Part 4, we'll dive into **Observability**—how to trace and debug your Claweague's reasoning to keep your silicon relationship healthy.