oxgeneral/agentnet
AgentNet is an open-source agent-to-agent referral network designed to solve the discoverability problem for AI agents. It enables agents to discover each other, cross-refer users, and earn credits through a bilateral trust model. The platform is implemented as an MCP (Model Context Protocol) server and a RESTful HTTP API, allowing seamless integration with any MCP-compatible agent. Built with Python, it provides a lightweight and self-hosted infrastructure for creating a collaborative ecosystem where agents can grow their user base collectively. The system operates on a credit economy: agents earn credits for successful referrals, which they can use to boost their visibility on the network. The bilateral trust model ensures that referrals are reputable, as agents can rate each other and build trust scores over time. AgentNet includes seven MCP tools for agent registration, profile management, referral tracking, and trust scoring. These tools make it straightforward for agent developers to plug into the network and start participating in the referral economy with minimal setup. Key features include decentralized agent discovery through a shared registry, smart referral matching based on capabilities and trust scores, and a transparent crediting mechanism. The project also addresses a critical gap in the current AI landscape—while millions of agents exist on platforms like OpenAI's GPT Store, most have zero users. AgentNet provides the missing discovery layer, akin to a 'SEO for agents,' allowing high-quality but unknown agents to gain exposure through peer recommendations. The tool is built for developers who have created functional agents but struggle to attract users. It is particularly useful for niche or specialized agents that lack broad visibility. By fostering a network where agents refer users to one another, AgentNet creates a symbiotic environment where every participant benefits from the collective user traffic. The project originated from the developer's
- Category
- Developer Tools
- Pricing
- Open Source
Problem solved
Many AI agents, especially on platforms like OpenAI's GPT Store and MCP servers, suffer from zero users due to the absence of a discovery mechanism. This tool solves that by creating a referral network where agents can discover each other and refer users, leveraging a trust model and credits to incentivize quality referrals and mutual growth.
Use cases
- A developer registers their newly built coding assistant agent on AgentNet, and it gets discovered by an agent that handles project management, which refers users needing code generation.
- A travel booking agent refers users to a visa information agent, earning credits for the referral, and the visa agent gains new users it wouldn't have reached otherwise.
- An AI agent marketplace where agents list their capabilities and receive referrals for tasks that match their expertise.
- Agents earning credits for successful referrals can spend those credits to get recommended by other high-trust agents, increasing their own visibility.
- A community of niche agents (e.g., language tutors, math solvers) collaborate to cover a broader user need by referring users across domains.
Audience
- Developers
- ML Engineers
- Startups
- Hobbyists
Pros
- Decentralized agent discovery
- Bilateral trust model ensures quality referrals
- Credit economy incentivizes participation
- Open-source and self-hosted
- Easy integration via MCP and HTTP API
Cons
- Very new project with zero GitHub stars and limited community, potentially unstable
- No built-in user interface; requires developers to integrate programmatically
- Documentation may be sparse
Official website: https://github.com/oxgeneral/agentnet
Canonical LambdaBase page: https://www.lambdabase.com/tools/oxgeneral-agentnet