What problem does it solve? Agents need structured, reusable tools for complex operations, but embedding large code blocks in chat context is messy and untestable. This Skill defines how to implement agent-callable actions as standalone TypeScript files that the agent invokes with pnpm action <name>. ## Core Features & Use Cases - Action scaffolding: Create actions in actions/ using parseArgs(), loadEnv(), fail(), and agentChat.submit() from @agent-native/core. - Schema-validated actions: Use defineAction with a Zod schema for runtime validation, typed run() arguments, and auto-generated JSON Schema for the agent's tool definition. - Common patterns: Includes templates for API integration actions (e.g., image generation) and data processing actions that read/write JSON files. - Use Case: You need the agent to generate an image from a prompt. Create actions/generate-image.ts, then the agent runs pnpm action generate-image --prompt "..." --output data/image.png and reports the result back to chat. ## Quick Start Ask the agent to create a new action that reads a JSON input file, transforms the data, and writes the result to an output path, then run it with pnpm action.
