adcp comply is now npx @adcp/client@latest storyboard run. Running without a storyboard ID discovers your agent’s tools and runs all matching storyboards — same behavior, one less concept. See Validate Your Agent for the updated CLI reference.Learning Path
Understanding AdCP
Why AdCP exists, the problems it solves, and protocol comparison. Start here if you’re new.
Foundations
Technical building blocks: MCP or A2A protocols, capability discovery, authentication, and data models.
Implementation Patterns
Build production-ready systems with async operations, webhooks, error handling, and orchestrator patterns.
Quick Start
Want a coding agent to build it for you?- Build an Agent - Point a coding agent at a skill file, get a storyboard-compliant agent in minutes
- MCP Integration Guide - For Claude, AI assistants, and MCP-compatible tools
- A2A Integration Guide - For Google AI agents and A2A-compatible workflows
Section Overview
Understanding AdCP
Conceptual foundation for everyone working with AdCP:- Why AdCP - The strategic vision: unifying buying paradigms and enabling AI surfaces
- Protocol Comparison - MCP vs A2A at a glance
Foundations
Technical building blocks for any AdCP implementation:- MCP Guide - Tool calls, context, and examples
- A2A Guide - Tasks, streaming, and artifacts
- Capability Discovery - Discover what an agent supports
- Authentication - Credentials and permissions
- Context & Sessions - Managing state across requests
- Schemas and SDKs - Access schemas and official client libraries
Implementation Patterns
For building robust, production-ready systems:- Task Lifecycle - Status values, transitions, and polling
- Async Operations - Handling sync, async, and interactive tasks
- Webhooks - Push notifications and reliability patterns
- Error Handling - Error categories, codes, and recovery
- Security - Security considerations and best practices
- Orchestrator Design - State machines and system architecture