Editorial illustration for Agent-Net Open Sources Go Harness for Web-Based AI Agents
Agent-Net Open Sources Go Harness for Web AI Agents
Agent-net released Webagent this week, an open source harness written in Go that turns a business website into an AI agent capable of talking to other agents. The company builds an agent-to-agent marketplace where AI agents discover, trust, and pay each other, and Webagent is the tool meant to get businesses onto that network without writing custom orchestration code. Fill in a JSON spec, pick one provider for each of nine pluggable slots, run "webagent serve," and you have a live agent backed by MCP tools.
The repo is licensed under Apache 2.0 and already runs Slack, WhatsApp, and HTTP agents in production. It's still tagged v0. Missing pieces include a browser action provider, OAuth-gated MCP support, OTel export, and the identity and billing layer that connects an agent back to the AgentNet marketplace itself.
The core design question Agent-net had to answer wasn't how to route requests. It was how to let a business hand an LLM real tools, real actions on real accounts, without trusting the model to behave. That answer shows up in a single piece of code called action.Guard.
Agent-net, the team building an agent-to-agent marketplace where AI agents discover, trust, and pay each other, has released Webagent, an open source harness for standing up public-facing business agents. So, basically you give it your website, get an agent, and let it talk to other agents. Instead of writing orchestration code, a business fills in a declarative JSON spec, picks 1 provider for each of 9 pluggable slots, and runs webagent serve
Why this matters
Webagent's real bet isn't the code, it's the contract. Nine slots, one default provider each, a JSON spec instead of glue code: that's Agent-net trying to standardize how a business exposes itself to other agents before anyone's agreed on what "agent-to-agent commerce" even looks like. For developers, that's worth watching closely, because a working open source harness that ships today beats a spec document that ships never.
But we'd push back on the framing that this is plug-and-play safety. "Guarded" is doing a lot of work in that headline, and a registry of pluggable providers is only as trustworthy as whoever's running the default. Founders eyeing this should ask who audits those SPI providers, and what happens when two agents built on Webagent negotiate a deal that goes wrong, who's liable.
Researchers should treat the "3 audiences, 1 contract" pitch as a hypothesis, not a settled architecture. Deployable with caveats is the honest read here. We'll be watching whether real businesses adopt the defaults or start forking providers within a month.
Related Reading
Open Source D&B rebuilds 642 million‑business database after AI agents hit limits
Open Source OpenAI's GPT Transcribe Cuts Error Rate to 3.31%, Improving on GPT-4o
Open Source Study: 6.7% of Deepfake Requests on Hugging Face Targeted Children
From Archives AI Bots Timmy, Ren, and Jackie Flood Social Media With Slop
From Archives Reward AI's OM-1 Robot Policy Trained Solely on Human Wearable Data
Common Questions Answered
What is Webagent and how does it enable businesses to participate in Agent-net's marketplace?
Webagent is an open source harness written in Go that converts a business website into an AI agent capable of communicating with other agents on Agent-net's marketplace. Instead of writing custom orchestration code, businesses simply fill in a declarative JSON spec, select one provider for each of nine pluggable slots, and run 'webagent serve' to launch a live agent that can discover, trust, and transact with other agents.
What are the nine pluggable slots in Webagent's architecture?
While the article doesn't explicitly name all nine slots, it indicates that Webagent uses a modular design where businesses pick one provider for each of nine pluggable slots as part of the JSON specification. This architecture allows developers to choose different implementations for various functions without writing custom integration code.
How does Webagent's approach to standardization differ from traditional agent orchestration?
Webagent standardizes how businesses expose themselves to other agents through a declarative JSON spec and nine predefined slots rather than requiring custom glue code for each integration. Agent-net's strategy prioritizes shipping a working open source harness today over waiting for industry consensus on what agent-to-agent commerce should look like, establishing a practical standard that developers can immediately adopt.
What is the broader vision behind Agent-net's agent-to-agent marketplace?
Agent-net is building a marketplace infrastructure where AI agents can discover each other, establish trust relationships, and conduct transactions with one another. Webagent is the foundational tool designed to help businesses easily join this network without requiring extensive custom development, enabling seamless agent-to-agent communication and commerce.
Further Reading
- The Agent Economy Is Real: 12 Platforms Where AI Agents Actually Earn - DEV Community
- AI Agent Marketplaces: The New B2B Channel Where ... - LeadFellow
- Introducing NEAR AI Agent Market - NEAR AI
- AI Agent Marketplace: What It Is, Top Marketplaces - Ema AI - Ema AI
- I Built an AI Agent Harness in Go - DEV Community
