Discover and explore top open-source AI tools and projects—updated daily.
agent-network-protocolAn SDK for building a collaborative network of intelligent agents
Top 95.8% on SourcePulse
<2-3 sentences summarising what the project addresses and solves, the target audience, and the benefit.> Agent Network Protocol (ANP) addresses the need for standardized, secure, and efficient communication among intelligent agents. AgentConnect, its open-source SDK, enables agents to form collaborative networks, aiming to become the foundational communication layer for the "Intelligent Agent Internet Era." It benefits developers by providing robust tools for agent interoperability and network formation.
How It Works
The project implements the ANP standard through two primary SDKs: OpenANP and ANP Crawler. OpenANP offers a decorator-driven approach for building agents, automatically generating necessary descriptive documents (ad.json, interface.json) and JSON-RPC endpoints. It simplifies agent creation and integrates LLM capabilities via OpenAI Tools format export. ANP Crawler provides a fetch-and-parse mechanism, akin to a web crawler, for ANP documents, ideal for LLM tool integration and deterministic data collection.
Quick Start & Requirements
Installation is available via pip (pip install anp) or source compilation using uv sync after cloning the repository. Core dependencies include Python and FastAPI. Examples for various modules like OpenANP, ANP Crawler, and DID-WBA authentication are readily available within the repository. Key documentation links are provided within the README.
Highlighted Details
ad.json) and OpenRPC interface (interface.json) documents automatically.Maintenance & Community
The project is authored by GaoWei Chang. Community engagement is facilitated through a Discord server. The source code is hosted on GitHub.
Licensing & Compatibility
The project is licensed under the permissive MIT License, allowing for broad compatibility with commercial and closed-source applications.
Limitations & Caveats
The README does not explicitly detail limitations. The ANP Crawler's type safety is described as "Dict-based returns," which may require additional handling for strict type checking compared to the proxy-style RemoteAgent.
2 days ago
Inactive
ArcadeAI
agi-inc
letta-ai