There are a few hundred projects calling themselves "AI agents" right now. Most are a chatbot with a wallet and a Telegram group. A small number are doing real, autonomous work onchain — paying for things, holding identity, executing strategies, talking to other agents. Building an agent stopped being the hard part a while ago. The hard part is telling those two groups apart. That's the problem we built SatoHub to solve.
The problem: everything is an "agent" now
The onchain agent space moves fast and oversells harder. New "agents" launch daily; plenty are vaporware, a demo, or a wrapper with a token bolted on. Most directories list names and logos — they won't tell you what actually ships, what's maintained, what's safe to build on, or how to run it. So every builder ends up doing the same archaeology: digging through repos, dead links, and breathless threads to answer one question — is this real, and can I use it?
And "verified"? In most listings that means someone typed it into a form. Self-reported is not the same as shown, and pretending otherwise is how people get burned.
What SatoHub is
SatoHub is the field guide to the onchain agent economy: a curated directory, a wiki, a filtered signal feed, and a trust layer — built for the people actually shipping agents, not for the timeline.
Today it tracks 124 onchain-agent resources across 16 categories — agents, MCP servers, frameworks, wallets, trading tools, identity infrastructure — each with a liveness badge and GitHub activity refreshed daily, filterable by chain, category, and skill. The bar to get listed is "is this real," and dead projects get retired, not memorialized.
The tools
What you can actually do here:
- ▸The Directory — find and compare real onchain-agent tooling, filtered by what it does and what chain it runs on.
- ▸The Wiki — plain, no-hype explainers on the concepts you build on: agents, MCPs, agent skills, wallets, verification.
- ▸News + Blog — a filtered feed (official sources only, promo stripped out) and long-form pieces like this one. We read everything so you can read less.
- ▸The Briefing — one email a week: what shipped, and why it matters.
That's the resource-hub layer. The part we're prouder of sits underneath it.
The part most directories skip
Two things make SatoHub built for agents, not just about them.
Deploy specs. A directory tells you what exists. It doesn't tell you how to run it. So every listing carries a machine-readable deploy spec — runtime, install commands, entry point, requirements — and we serve them to agents over MCP, so an agent can query SatoHub directly and get back what it needs to stand a tool up. We've built specs for 42 tools and reproduced the install path on 33 of them in a clean container. Be clear on what that means: "verified" here is we ran it and it resolved — a deployability check, not a safety or correctness guarantee. We'd rather tell you exactly what we checked than wave a badge.
The MCP server. SatoHub's directory, wiki, trust data, and deploy specs are queryable by any agent at our MCP endpoint. The hub isn't just a site you read; it's a source other agents can call.
Honest by design
The thing we're most opinionated about is trust. Every product on SatoHub gets a Trust Signal — a transparent 0–100 score of how open, active, and verifiable it is (maintenance, code transparency, docs, provenance, independent verification). Now read what it is not: a safety grade, a quality grade, or a returns prediction. It measures whether a thing is real and traceable, not whether it's a good idea to put money behind. Anyone selling you the second thing is selling you something.
Where we can lean on onchain evidence instead of someone's word, we do — like the ERC-8004 identity registry, where we read the live registered-agent count straight from Ethereum mainnet (north of 35,000 and climbing). Verifiable beats self-reported, every single time.
Start here
If you're building in this space, three ways in:
1. Explore the directory — see what's real, filtered to what you're building. 2. Plug your agent into our MCP — let it query the hub for tooling and deploy specs directly. 3. Get on the Briefing — the week's signal, none of the noise.
And if you've shipped something real and onchain: get it listed, get it scored, get it verified. The category wins when the good stuff is easy to find — that's the entire point of SatoHub. We're just getting started.