a2a-bridge CLAUDE.md

Project instructions for a2a-bridge, a two-way connection between Claude Code and other AI coding agents.

In plain words
What is it for?
Developing or reviewing peer-agent adapters, multi-machine setups, concurrent sessions, and repository documentation.
Why use it?
They keep repository content in English and set rules for the project's architecture, supported deployments, and adapters.

Instructions file

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add instructions/firstintent/a2a-bridge/claude-md
Clone the repo
git clone --depth 1 https://github.com/firstintent/a2a-bridge
Per session 1,740 This file is loaded in full into every session.
When invoked 1,740 The same file — it is already loaded in full.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.01740 $0.01740
Opus 5 $0.00870 $0.00870
Sonnet 5 $0.00348 $0.00348
Haiku 4.5 $0.00174 $0.00174

Measured 2d ago against content hash af1239ec4093, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

a2a-bridge CLAUDE.md scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

CLAUDE.md · 167 lines

How it starts

The opening of the file, as written. The whole thing — 167 lines — stays where its author put it; the contents beside it link to each section on GitHub.

CLAUDE.md — a2a-bridge

Project

a2a-bridge is a bidirectional bridge between Claude Code and other AI coding agents (Codex, OpenClaw, Hermes, ...) built on the Claude Code Channels protocol. Modelled after raysonmeng/agent-bridge but designed from day one for:

  • Pluggable peer adapters (one adapter per target agent)
  • Multi-machine deployment (daemon on one host, CC on another)
  • Multi-room / concurrent CC sessions

Language policy — strict

Everything stored in this git repo MUST be in English:

  • Source code, including variable names and string literals
  • All code comments
  • Commit messages (subject and body)
  • Pull request titles and descriptions
  • Issue titles and bodies
  • File and directory names
  • README, ARCHITECTURE, ROADMAP, and any other top-level docs

English-language documentation targeted at readers of the repo lives under docs/ and is committed (cookbook, runbooks, subsystem guides). Chinese-language scratchpads — local design notes, meeting memos, translation drafts — live under docs-cn/, which is git-ignored (see .gitignore) and must never be committed.

Rationale: this is an open-source project targeting a global audience. Mixed-language repos fragment the reader base and confuse grep, so the committed tree stays English-only; docs-cn/ exists to keep the local workflow useful without polluting the repo.

Stack

  • TypeScript, Bun runtime (matches upstream agent-bridge)
  • MIT license
  • Two-process topology: foreground MCP plugin + persistent daemon
  • Peer adapters implement a common IPeerAdapter interface

Architecture constraints — enforced

The src/ tree is organized by runtime + domain, not by abstract "layer number":

src/
  shared/          zero-dep utilities (state-dir, daemon-lifecycle, config-service, logger)
  messages/        shared value objects (BridgeMessage, ...)
  transport/       plugin <-> daemon control plane
  runtime-plugin/  code that runs inside the Claude Code MCP plugin
    claude-channel/    MCP channel adapter
    daemon-client/     client side of the control plane
    bridge.ts          plugin process entrypoint
  runtime-daemon/  code that runs inside the persistent daemon
    peers/             outbound peer adapters (codex, openclaw, hermes)
    inbound/           A2A server exposing Claude Code to external clients
    rooms/             RoomRouter
    tasks/             Task lifecycle + storage
    daemon.ts          daemon process entrypoint
  cli/             user-facing CLI; composition root for integration tests

Read the full file on GitHub · 167 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 167 lines · 1,740 tokens per session scan A af1239ec4093

Subscribe to this mod's changes

a2a-bridge CLAUDE.md is an instructions file published in the GitHub repository firstintent/a2a-bridge (8 stars, last pushed 4mo ago), licensed MIT. It adds 1,740 tokens to every session, about $0.0087 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.