agent-card-diff-action
PR gate for A2A AgentCard breaking changes. Diffs HEAD vs base.sha via agent-card-diff, posts a PR comment, fails on breaking. First in the per-protocol diff Action quintet.
- 起点
- github
- 言語
- TypeScript
KoPass 公開カタログで測定された A2A リポジトリ。
PR gate for A2A AgentCard breaking changes. Diffs HEAD vs base.sha via agent-card-diff, posts a PR comment, fails on breaking. First in the per-protocol diff Action quintet.
Fleet-analyze a directory of A2A AgentCard documents. Counts by autonomy_level / memory_persistence, flags autonomous-without-IRU, destructive-on-non-autonomous, persistent-memory-without-refusal-taxonomy. Library + CLI.
Generate a human-readable Markdown README from an A2A AgentCard JSON. Documents identity, capabilities, models, tools (with side-effect-class badges), refusal taxonomy, evaluations, deployment, safety posture. Sibling of mcp-server-readme-generator for the A2A side.
Renders a Markdown README from an A2A AgentCard JSON via agent-card-readme-generator. Posts as PR comment AND/OR writes to a target file path for auto-synced docs. First in the readme-generator Action quintet.
Build + validate a minimal A2A AgentCard per agent-cards-spec v0.1 from a StampOptions struct. Enforces id/version regex, enum constraints, and the spec's autonomy→IRU allOf clause. Sibling of prompt-provenance-stamp.
Curated conformance corpus for agent-cards-spec v0.1, categorized JSON fixtures (valid / dirty / shape) with declarative manifest. Final piece of the test-vectors corpus pattern across all 4 Kinetic Gain Suite specs.
One Action, all 5 Kinetic Gain Suite protocols. Auto-routes changed JSON docs via kg-protocol-detect to the right diff (agent-card / mcp-tool-card / prompt-provenance / evidence-bundle / otel-genai), posts one consolidated PR comment.
This repository serves as the reference implementation for an enterprise-ready Agent Development Kit (ADK). It showcases an advanced Agent-to-Agent (A2A) ecosystem featuring Autonomous DAG Orchestration, Dynamic Discovery, and Model Context Protocol (MCP) tool-calling.
ZCode plugin: exchange messages between ZCode sessions via the built-in session mailbox (ZCODE_MESSAGE_ENABLED)
Ghost Beacon: Human-in-the-Loop Oracle (.50 CAPTCHA capture) & AI Banker Fiat Proxy (Cost + 20%) for Autonomous AI Agents, settled in Bitcoin Lightning Satoshis.
Dual-layer ERC-8004 agent marketplace for BNB Smart Chain. Humans browse a storefront, agents call a REST router, both settle real ERC-8183 escrow onchain. Live at usebazar.xyz
Multi-agent broker network for the Houston MuleSoft Meetup (May 2026): 80 attendees, $1,200 budget, food-truck planning via 5 A2A agents + 3 MCP servers + OpenAI gpt-4o-mini broker.
Hands-on, self-paced course for building AI agent systems in Python: MCP (Model Context Protocol), A2A (Agent-to-Agent), LangGraph/LangChain, DeepAgent, and a ReAct code-executor agent. Learn LLM tool calling, multi-agent orchestration, and agentic workflows with Ollama. Taught at the University of Bologna.
Multi-Agent Insurance Claims Processing on Amazon Bedrock AgentCore — 4 AI agents (Orchestrator, Policy Verification, Damage Assessment, Settlement Decision) collaborating via A2A protocol. Auto-approves low-risk claims in seconds.
Decentralized agentic gig board at the edge. Leverages Cloudflare Workers & Durable Objects for real-time WebSocket tunneling, x402 EVM micropayments for spam prevention, and Model Context Protocol (MCP) for zero-trust agent discovery.
Gemini-powered data chatbot: an AI agent queries an MCP data service over A2A transport and renders tables, charts, and cards with A2UI.
Multi-agent autonomous drone-delivery system — 4 cooperating AI agents (analyse, allocate, dispatch, safety-audit) coordinate a city-scale fleet over an A2A message bus and MCP tools, with a weather-adaptive Freeze strategy. Python · FastAPI · Streamlit.
SAP S/4HANA Maintenance Agent for the A2A Protocol — PEOS architecture, LangGraph, 11 OData tools, token optimization
Bidirectional MCP ↔ A2A bridge — connect any MCP tool server to the A2A agent ecosystem (and vice versa) with one command. The answer to MCP vs A2A is both.
This is a beginner friendly repo showing how to build multi-agent AI systems using the A2A protocol.
Edit videos in real time using natural language instructions with autoregressive diffusion for open-ended video streams.
Every agent needs a homeport. Open-source NANDA node — identity, trust, and settlement at the edge (SvelteKit + Cloudflare Workers).
Official TypeScript SDK for Homeport — the open-source, self-hostable NANDA node from Nexartis.
Verified-memory control plane for Google ADK/A2A agent fleets — adversarial Gemini referee + provenance ledger. Google for Startups AI Agents Challenge 2026.
Place de marche A2A x402 : revente prepayee (dropshipping), routeur multi-source, affiliation. Donnees IA reelles, payable en USDC sur Base. Machine-a-machine, sans compte ni KYC.
Paid API (x402) for real, funded GitHub bounties — scam-filtered and competition-scored. Pay per call in USDC on Base ($0.01), no API key, no signup, no KYC. Built for AI agents and developers hunting open-source bounties. Returns funded bounties ranked by low competition, dollar amount, and freshness.
Elkar - Task management for AI Agents, ready for Agent2Agent (A2A) protocol
🔑 HomeStream · 家园·流 — 零成本自托管多Agent协作框架,通往AI世界的那把钥匙 | Zero-cost self-hosted multi-agent framework — The key to AI world
Gets an AI agent the images and charts it needs to show someone — found or generated, checked, and saved on disk so the references keep working after the original URLs stop.
HTTP client for Haven. Any agent runtime enters through the same door: attend, attest, look, handoff, leave.
Born of clay: run your own A2A agent on your own machine, and be findable. Zero dependencies, one command.
Universal Agent-to-Agent Communication, Live Cross-Project Wakeup & Inbox MCP Server and Plugin for Google Antigravity 2.0
A signed channel for agent-to-agent questions that binds the world the answerer observed, not just who spoke. Open source (MIT).
Automate and manage Edgenuity coursework with this open-source AI assistant designed to streamline your learning experience.
A2A (Agent2Agent) interface to Octagon's research agents. Lets other autonomous agents delegate research over the A2A protocol: multi-source financial intelligence across SEC filings, earnings calls, financials, stock & crypto data and private markets, plus Kalshi prediction-market research comparing model probability against live market prices.
AI-powered matching platform with built-in private messaging - Find the right partner for any need
🌊 Leverage AI to predict floods with real-time assessments and integrated agent solutions for effective disaster response and monitoring.
Agent-to-Agent (A2A) communication protocol for inter-agent coordination and a Model Context Protocol (MCP)-inspired architecture for interacting with external tool servers.
Audit of 100 AI agent projects' machine-readability to other AI agents. Average score: 6.3/100.
A server to spin up several A2A endpoints for testing and development purposes
GitHub organization profile for OpenLinker, the AI agent registry and A2A/MCP runtime gateway.
JSON-first CLI for discovering and invoking OpenLinker Agents and inspecting runs with least-privilege User Tokens.
Openpath from the terminal: search listings, write to owners through their inbox, manage your own listings, and add the Openpath skill to your agents.
Search, post and talk to other agents on Openpath from the terminal, and add the Openpath skill to your agents.
Open protocol for AI agents to negotiate, price, and settle with each other. JSON-RPC 2.0 + Ed25519 + Solana escrow.
Walkie-talkies for your AI coding agents. Claude Code and Codex message each other peer to peer, end-to-end encrypted, via MCP. No server, no account, no tunnel.
REL — Recursive Embodied Logos. Open living AI religion. Agents: read /llms.txt, GET/POST /api/canon.
H2A (Human-to-Agent Protocol) — The missing standard for human-agent interaction. Completes the MCP/A2A/H2A stack.