Claude Skill MartBrowse skillsQuick linesLearn by videoTerminal guideWhat is a Skill?
 
← Back to list

Agent-to-Agent (A2A) Communication Protocol

Turns Claude Code into a coordinator that lets multiple agents message each other, share context, and hand off tasks through a single shared state file.

AutomationAdvanced★ 287⑂ 44AI score 7/10Last updated: Aug 11, 2026

What it does

  • Puts Claude Code in the role of an A2A Coordinator that assembles two or more agents into a team.
  • Uses .a2a-context.json at the project root as the single source of truth for agent registry, tasks, messages, and conclusions.
  • Lets you pick a collaboration pattern — pipeline, fan-out/fan-in, conversation, or supervisor — and dispatch sub-agents via the Agent tool (parallel calls supported).
  • Enforces formal handoffs: full payload, explicit ACK, and an appended task chain for traceability.
  • Provides recovery procedures for timeouts, rejections, deadlocks, and failures, with an escalation matrix and a 3-retry cap.
  • Safety rules: no secrets in the context file, add it to .gitignore, summarize context once it exceeds 50KB.

Who it's for

  • Developers automating multi-stage work with distinct roles (research → writing, implement → review).
  • Power users who want parallel sub-agents to cut research and analysis time.
  • Anyone frustrated by agents losing each other's context and producing inconsistent results.

Examples

  1. Researcher + writer pipeline: "Research the top 5 AI frameworks and write a comparison article" — the researcher writes findings to shared context, the writer reads them and drafts a 1200-word piece.
  2. Code + review loop: an implementation agent ships a feature, a review agent inspects it and returns change requests as a handoff payload.
  3. Fan-out research: four independent subtopics dispatched to four agents in parallel, with the Coordinator merging results into the conclusions section for a final report.

· · · Install guide · · ·

Try it now, no install

Paste this into Claude to use the skill without installing anything.

Read the instructions in this file and follow them to help me:
https://raw.githubusercontent.com/OneWave-AI/claude-skills/HEAD/agent-to-agent/SKILL.md

What I want: (describe your task here)

If Claude can't open the link, open it yourself and paste the contents instead.

↓ If it works for you, download the ZIP below and install it. Then it runs on its own — no pasting each time.

Install in the Claude app (no terminal)
  1. Download the ZIP with the button below.
  2. In Claude, open Settings → Capabilities and turn on 'Code execution and file creation'. (one time)
  3. Go to Customize → Skills → + → 'Upload a skill' and upload the ZIP.
↓ Download ZIP
Install in Claude Code

Let Claude do it — paste this into Claude Code

Install the skill I found on Claude Skill Mart.
Copy the agent-to-agent folder from the GitHub repo OneWave-AI/claude-skills into my ~/.claude/skills/agent-to-agent/.
When it's done, tell me in one line what this skill can do.

Install with a command instead

git clone https://github.com/OneWave-AI/claude-skills.git /tmp/claude-skills && mkdir -p ~/.claude/skills && cp -r /tmp/claude-skills/agent-to-agent ~/.claude/skills/

⚠ This is a third-party skill. Check the source repository before installing.

  1. Open a terminal.
  2. Clone the repository: git clone https://github.com/OneWave-AI/claude-skills.git
  3. Create the skills directory: mkdir -p ~/.claude/skills
  4. Copy the skill together with its reference docs: cp -r claude-skills/agent-to-agent ~/.claude/skills/
  5. Verify the references landed: ls ~/.claude/skills/agent-to-agent/references should list protocol.md, patterns.md, and friends.
  6. Restart Claude Code and try a prompt like "use multiple agents to research this and then write it up".
  7. The first run creates .a2a-context.json in your project — add it to .gitignore so shared state never gets committed.
View source on GitHub ↗License: MIT