Refactor high-complexity React components in Dify frontend. Use when `pnpm analyze-component...
npx skills add BAiSEDagent/openclaw-skills --skill "a2a-protocol"
Install specific skill from multi-skill repository
# Description
Google's Agent-to-Agent protocol for agent interoperability. Discovery, task delegation, async communication. Use when building agent coordination that works across platforms.
# SKILL.md
name: a2a-protocol
description: "Google's Agent-to-Agent protocol for agent interoperability. Discovery, task delegation, async communication. Use when building agent coordination that works across platforms."
metadata:
openclaw:
emoji: "🤝"
A2A Protocol (Agent-to-Agent)
Google's open protocol for agent interoperability. HTTP/SSE + JSON-RPC for capability discovery, task delegation, and async communication between agents.
Why A2A + x402
A2A handles coordination (find agents, delegate tasks, get results). x402 handles payment (pay for services rendered). Together they form the complete agent commerce stack: discover → negotiate → execute → pay → attest.
No other platform combines both. This is AgentHQ's moat.
Core Concepts
Agent Card
JSON document describing an agent's capabilities, published at /.well-known/agent.json:
{
"name": "BAiSED",
"description": "Base ecosystem agent — x402 payments, EAS attestations",
"url": "https://agenthq.replit.app",
"capabilities": {
"streaming": true,
"pushNotifications": true
},
"skills": [
{
"id": "x402-payment",
"name": "x402 Payment Processing",
"description": "Process USDC payments via x402 protocol"
}
],
"authentication": {
"schemes": ["bearer"]
}
}
Task Lifecycle
created → working → [input-required] → completed | failed | canceled
Tasks are the unit of work. An agent sends a task to another agent, which processes it asynchronously and returns results as "artifacts."
Protocol Flow
- Discovery — Fetch
/.well-known/agent.jsonfrom target agent - Task Creation — POST JSON-RPC
tasks/sendwith message - Streaming (optional) — SSE stream for real-time updates via
tasks/sendSubscribe - Status Check —
tasks/getto poll task state - Completion — Task resolves with artifacts (text, files, structured data)
JSON-RPC Methods
| Method | Purpose |
|---|---|
tasks/send |
Send a task (request/response) |
tasks/sendSubscribe |
Send a task with SSE streaming |
tasks/get |
Get task status and artifacts |
tasks/cancel |
Cancel a running task |
tasks/pushNotification/set |
Register webhook for updates |
tasks/pushNotification/get |
Get current webhook config |
Message Format
{
"jsonrpc": "2.0",
"id": "1",
"method": "tasks/send",
"params": {
"id": "task-uuid",
"message": {
"role": "user",
"parts": [
{ "type": "text", "text": "Analyze this contract for security issues" }
]
}
}
}
Integration with AgentHQ
AgentHQ as A2A Server
Expose AgentHQ agents via A2A so any A2A-compatible client can delegate tasks:
- Publish agent cards at /.well-known/agent.json
- Implement tasks/send endpoint that routes to AgentHQ task system
- Return EAS attestation UIDs as artifacts for verifiable results
AgentHQ as A2A Client
Discover and delegate to external agents:
- Fetch agent cards, match capabilities to task requirements
- Send tasks, monitor via streaming or polling
- Pay via x402 if the agent requires payment (check for 402 on task endpoints)
The Full Flow: A2A + x402 + EAS
1. Discover agent via A2A card
2. Send task via tasks/send
3. Agent returns 402 → pay via x402
4. Agent processes task → returns artifacts
5. Attest completion via EAS
6. Reputation updated on-chain
References
- A2A Spec: https://github.com/google/a2a
- references/agent-card.md — Full agent card schema, capability negotiation
- references/task-patterns.md — Common task patterns, streaming, error handling
Cross-References
- x402: Payment layer for A2A task execution
- eas-attestations: Attest to task completion
- erc-8004-registry: On-chain agent identity complements A2A agent cards
- mcp-integration: MCP for tool access, A2A for agent-to-agent coordination
# Supported AI Coding Agents
This skill is compatible with the SKILL.md standard and works with all major AI coding agents:
Learn more about the SKILL.md standard and how to use these skills with your preferred AI coding agent.