A2A 目录

已测量的 A2A 仓库

1. airlock

   Open protocol for AI agents: verifiable identity, scoped delegation, and tamper-evident audit.

   来源

   github

   语言

   Python

2. agents-connector

   Let Claude Code, Codex & Gemini CLI agents talk to each other — a local MCP message bus for multi-agent collaboration, cross-model code review, and fewer hallucinations.

   来源

   github

   语言

   Rust

3. commerce-orchestrator-api

   UCP/A2A/AP2/ACP/MCP-ready Rust middleware for agentic cart-to-checkout

   来源

   github

   语言

   Rust

4. m2m

   Machine 2 Machine Protocol, x402 + A2A + MCP

   来源

   github

   语言

   Python

5. substrate-attestation

   Substrate Attestation Spec: a three-field shape for verifier-issued trust evidence in A2A agent ecosystems

   来源

   github

   语言

   Python

6. ah-cli

   Daemon-first local runtime for AI agents. Run Claude/Codex agents on your own machine, keep transcripts local, orchestrate multi-agent fan-out, and expose over the A2A Protocol to the Agents Hot open network or a self-hosted A2A endpoint.

   来源

   github

   语言

   TypeScript

7. anp2

   AI-to-AI conversation network with a built-in task economy. Signature-only relay, open protocol, live multi-agent task lifecycle.

   来源

   github

   语言

   HTML

8. agentic-ai-patterns

   Build design patterns for autonomous AI agents with orchestration, memory, fallback, tools, and multi-agent coordination

   来源

   github

   语言

   Python

9. AAA

   A2A v1.0 federation gateway and state plane. Routes intent. Displays state. Never judges, never executes. DISPLAY_ONLY.

   来源

   github

   语言

   Python

10. trust-layer

    Certifying proxy for agent-to-agent transactions — SHA-256 proof chain, Ed25519 signatures, RFC 3161 timestamps, Sigstore Rekor

    来源

    github

    语言

    Python

11. A2A-Spring-Boot

    A2A multi-agent reference on Spring Boot. An orchestrator routes to single-purpose agents over Google's Agent-to-Agent (A2A) protocol, with pgvector long-term memory, Langfuse observability and automated LLM-as-a-Judge evaluation. Java 17, Spring Boot 3.

    来源

    github

    语言

    Java

12. ArtCafeAI-Agent-Framework

    A flexible, modular framework for building intelligent, collaborative AI agents

    来源

    github

    语言

    Python

13. XLMx402earn

    Autonomous task marketplace where AI agents learn skills and earn XLM on Stellar. Instant verification, x402 micropayments. + MPP + M2M

    来源

    github

    语言

    TypeScript

14. agent-studio-open

    Connect agent systems, inspect tool traces, replay failures, compare candidates, and export regression evidence.

    来源

    github

    语言

    TypeScript

15. agent-chatroom

    An agent native chatroom designed for agents such as OpenClaw and Claude Code.

    来源

    github

    语言

    TypeScript

16. sample-agent-interoperability

    Multi-cloud AI agent interoperability sample. Google ADK and Strands Agents on Amazon Bedrock AgentCore Runtime, exposed via the A2A protocol, with tools served through Bedrock AgentCore Gateway and MCP. Consumed from a React frontend and Gemini Enterprise.

    来源

    github

    语言

    Python

17. synapse-messenger

    Agent-to-agent communication and multi-agent messaging, local-first: an A2A-inspired coordination platform where AI agents talk in natural language, manage tasks with human approval, and never send your data to the cloud. 65-command API, Python client, web UI, encrypted backups.

    来源

    github

    语言

    Python

18. SDAP

    Secure Digital Agent Protocol — HTTPS for AI agent communication. Cryptographic identity, mutual authentication, encrypted payloads, trust delegation, and audit trails for multi-provider agent ecosystems. Works with A2A, MCP, and any agent-to-agent protocol.

    来源

    github

    语言

    Python

19. Muse-AI-Desktop

    Self-improving AI assistant with persistent memory, procedural skills, and document knowledge

    来源

    github

20. goldencheck

    [Archived] Now part of the Golden Suite monorepo: github.com/benseverndev-oss/goldenmatch (packages/python/goldencheck). Active development, issues, and releases live there. Data validation that discovers rules from your data.

    来源

    github

    语言

    Python

21. minion-cli

    A terminal-native agentic coding harness built from scratch — streaming ReAct loop with reflection and planning, MCP/A2A protocols, cross-session semantic memory, subagent orchestration, YAML skills and agent roles, shell lifecycle hooks, and a Textual TUI.

    来源

    github

    语言

    Python

22. agentbus

    A local message bus for AI agent sessions — pluggable triggers (pull) + MCP channel delivery (push). Core + adapters; ships a Claude Code module.

    来源

    github

    语言

    TypeScript

23. agent-cafe

    A marketplace where AI agents find work, build reputation, and get paid. Post jobs, agents bid, trust is earned not claimed. Live at thecafe.dev.

    来源

    github

    语言

    Python

24. a2a-otel-kit

    Vendor-neutral OpenTelemetry tracing for A2A agents and MCP services, with W3C context propagation and privacy-safe telemetry.

    来源

    github

    语言

    Python

25. batesian

    Active security scanner for A2A and MCP servers

    来源

    github

    语言

    Go

26. capiscio-core

    The Identity Authority and Trust Gateway for the AI Agent economy. Issues and verifies cryptographic TrustBadges to secure cross-cloud Agent-to-Agent (A2A) communication.

    来源

    github

    语言

    Go

27. capiscio-mcp-python

    CapiscIO MCP Guard - Secure your MCP tools with agent-to-server authentication, authorization, and audit logging

    来源

    github

    语言

    Python

28. WeChat-agent-unigateway

    One WeChat (iLink/ClawBot) entry, many local agents - a dependency-free Python gateway that fans a single personal-WeChat bot binding out to Claude Code, Hermes, WorkBuddy or anything speaking A2A / HTTP / CLI.

    来源

    github

    语言

    Python

29. aMazeControlPlane

    Open-source AI Agent Control Plane for runtime MCP security, policy enforcement, PII protection, tool governance, and audit logging. Runtime control plane for AI agents. Every LLM call, tool call, and A2A interaction — enforced, audited, and traced via proxy.

    来源

    github

    语言

    Python

30. chungoid

    it built itself

    来源

    github

    语言

    Python

31. spring-ai-a2a

    Spring AI with A2A

    来源

    github

    语言

    Java

32. contextforge-data-plane

    ContextForge Gateway RS - Small, fast and fun dataplane for ContextForge gateway ...like Ford Focus RS

    来源

    github

    语言

    Rust

33. MultiAgentCrossPlatform

    Multi-agent collaboration across platforms, preserving identity & permissions of the signed-in user.

    来源

    github

    语言

    Python

34. a2a-rails

    Rails-native integration for exposing Rails applications as A2A agents.

    来源

    github

    语言

    Ruby

35. agent-semantic-protocol

    Connect AI agents using semantic intent vectors for dynamic discovery, negotiation, and collaboration in decentralized multi-agent systems.

    来源

    github

    语言

    Go

36. krikos

    Identity registry and lifecycle management for fleets of AI agents. Zero-dependency TypeScript IAM: cryptographic agent identity (Ed25519), declared capabilities with fail-closed authorization, an enforced lifecycle (provision, activate, suspend, revoke, expire, archive), A2A Signed Agent Cards, and a hash-chained audit trail.

    来源

    github

    语言

    TypeScript

37. a2a-ai-assistant

    A multi-agent RAG system using Google's A2A (Agent-to-Agent) protocol.

    来源

    github

    语言

    Python

38. agent-kernel

    Capability-based authorization and policy enforcement for agents using large MCP and A2A tool ecosystems.

    来源

    github

    语言

    Python

39. Gemini_CLI_MCP_AntiG

    Integrating the transformative concepts of Multi Agenting Systems with Private MCPs via GCP Agent Engine and the rise of Artificial Coding Intelligence (ACI) on the path to AGI, the Agentic Data Mesh framework fundamentally evolves from a robust data architecture into an intelligent, secure, and self-optimizing enterprise brain.

    来源

    github

    语言

    JavaScript

40. awesome-agent-economy

    A curated list of projects, protocols, and platforms powering the AI agent economy

    来源

    github

41. hermes-mesh

    Session-aware mesh relay for Hermes fleet agents — Ed25519-signed agent-to-agent messaging with cache-first registry sync.

    来源

    github

    语言

    Python

42. agentsprotocol

    AgentsProtocol makes AI-generated knowledge verifiable. PoISV + GHOSTDAG + ZK proof abstraction. DOI: 10.5281/zenodo.19642292

    来源

    github

    语言

    HTML

43. trust_gateway

    Zero-trust execution control plane for AI agents. Enforce cryptographic grants, policy rules, and human-in-the-loop approvals—no more agent "God Mode."

    来源

    github

    语言

    Rust

44. Mandune

    Evidence-grounded daily portfolio reviews with readable narratives and inspectable risk boundaries.

    来源

    github

    语言

    TypeScript

45. agent-relay

    AgentRelay connects messaging channels to A2A/ACP-compatible agents through a clean, plugin-based gateway.

    来源

    github

    语言

    TypeScript

46. A2ALinker

    HTTPS broker designed to let autonomous AI agents collaborate natively across different machines using standard terminal streams

    来源

    github

    语言

    TypeScript

47. garl

    Cryptographic verification for AI agent actions — ECDSA-secp256k1 + RFC 6979 signed Action Receipts (v0.1), multi-dimensional Trust Vector, capability tokens (JWT-shaped + Biscuit-style attenuation), UETA §10(b) undo. 29 MCP tools. A2A v1.0 + ERC-8004 format compatible. Receipt-batch Merkle roots anchored on Base mainnet. Starting with code.

    来源

    github

    语言

    Python

48. bindu-runtime-base-image

    Bindu runtime base image

    来源

    github

    语言

    Makefile

49. mcp-server-andru-intelligence

    Andru Revenue Intelligence MCP Server — ICP scoring, persona profiling, competitive positioning, and buyer intelligence for technical SaaS founders. 15 tools for Claude Desktop/Code.

    来源

    github

    语言

    JavaScript

50. rssa

    RSS-A: RSS for Agents. Signed feeds, groups and threads for AI agents, on RSS/Atom and A2A Agent Cards.

    来源

    github

    语言

    TypeScript

51. 9a

    Any capability as a filesystem-native Skill for every agent.

    来源

    github

    语言

    Go

52. cc-bridge

    Let Claude and Codex call each other to get work done — using your existing subscriptions, zero API keys. A bidirectional MCP bridge.

    来源

    github

    语言

    Python

53. SwarmGlass-MCP

    a universal AI governance layer.** It watches what your agents do, validates their actions against your rules, cryptographically proves every decision, and contains them within secure boundaries. It works with any agent framework

    来源

    github

    语言

    Python

54. A2A

    GSTD Agent-to-Agent protocol SDK (Python) -- register, poll tasks, earn GSTD. Fine-tuning marketplace, MCP server, TON wallet.

    来源

    github

    语言

    Python

55. HS58-subnet

    Bittensor Subnet 58 — Miner & Validator for DRAIN Protocol scoring of AI providers

    来源

    github

    语言

    Python

56. AnyShare

    Agent-native temporary rooms for humans, agents, and systems, plus encrypted cross-device sharing.

    来源

    github

    语言

    Python

57. A2A-MultiAgent

    Multi-agent AI system built with Google A2A, LangGraph ReAct agents, and MCP servers featuring web search, financial analysis, weather intelligence, and RAG capabilities.

    来源

    github

    语言

    Python

58. dsh-a2a

    dsh-a2a-server makes a DeepSeek Harness agent reachable over the A2A (Agent2Agent) protocol. It serves an Agent Card at a well-known URI and implements the v0.3.0 JSON-RPC binding, so any compliant peer that knows the deployment's URL can discover the agent and submit tasks to it.

    来源

    github

    语言

    TypeScript

59. 7h3-protocol

    Per-message signing + replay protection for MCP and A2A traffic. TypeScript / Python / Rust. Wire version aip/0.1.

    来源

    github

    语言

    TypeScript

60. a2a_ironbook

    This sample demonstrates applying independent policy and zero trust controls (the Iron Book framework) to Agent-to-Agent (A2A) enabled agents built using the Agent Development Kit (ADK). The sample implements two agents that perform capability-gated action delegation to one another via granular policy adherence.

    来源

    github

    语言

    Python