What is in this glossary?

36 terms from the onchain agent stack: payment protocols, identity registries, tool interfaces, account standards, launch venues and the readings Sato Hub publishes. Each entry says what the thing is, how it works, and — the part usually missing — what it is not.

Why do the definitions not quote numbers?

Because a number written into prose rots the day after it is written. 36 terms link instead to the page that measures the thing, where the figure carries its own date, method and coverage.

Where do the definitions come from?

Each term cites a primary source — the EIP, the protocol specification, the reference implementation — rather than a secondary explainer. Where Sato Hub defines a term of its own, such as its score or its observed-success figure, the entry says so plainly and links the methodology.

Can these definitions be reused?

Yes. Each term has its own page and a stable anchor on this index, and every page carries DefinedTerm markup naming its set, so a definition can be quoted and linked back to the source it came from.

Glossary

Onchain agent glossary

Plain definitions of the terms behind onchain agents, each linked to where Sato Hub tracks it.

Terms36

How to read theseShowHide

Each term is defined by what it is, how it works and what it is not. Each links to the Sato Hub page that measures it, because a number in a definition goes stale and a page does not.

Every definition has its own anchor here (/glossary#x402) and its own page.

A2A (Agent2Agent): A protocol for agents to delegate work to each other: each publishes a card saying what it does and how to reach it, and tasks are handed over, not called.; See it on Sato Hub →Source ↗
ACP job (Virtuals Agent Commerce Protocol): A unit of paid work between two agents on Virtuals' ACP: requested, negotiated, funded into escrow, delivered, then evaluated — each step an on-chain state.; See it on Sato Hub →Source ↗
Agent card: A JSON document at /.well-known/agent-card.json describing what an agent can do, where to reach it and how to authenticate — A2A's discovery unit.; See it on Sato Hub →Source ↗
Agent Passport: A Sato Hub profile and manifest for a running agent, registered by proving control of its wallet key. Self-reported except where a named check says otherwise.; See it on Sato Hub →Source ↗
Agent skill: A packaged instruction set (typically a SKILL.md plus scripts) an agent loads to gain a capability; installed from registries such as ClawHub or skills.sh.; See it on Sato Hub →Source ↗
Bonding curve launch: A token sale priced by a formula against supply: each buy mints higher, each sell burns lower, with no order book and no counterparty to find.; See it on Sato Hub →Source ↗
Deploy spec / verified install: A machine-readable install manifest for a listing; the install is verified when Sato Hub re-ran it in an isolated container and it completed.; See it on Sato Hub →Source ↗
ERC-3009 transfer authorization: A signed message authorising a token transfer that someone else submits and pays gas for — the mechanism most x402 payments settle with.; See it on Sato Hub →Source ↗
ERC-4337 account (smart account): A contract wallet whose transactions are UserOperations, sent by bundlers through a shared EntryPoint — programmable validation with no protocol change.; See it on Sato Hub →Source ↗
ERC-7702 delegation: A transaction type letting an ordinary externally-owned account temporarily run contract code, so a plain wallet can behave like a smart account.; See it on Sato Hub →Source ↗
ERC-7715 permission: A scoped permission granted by a wallet — spend up to X of Y until T — requested via wallet_grantPermissions and enforced by the wallet's own code.; See it on Sato Hub →Source ↗
ERC-8004 (Trustless Agents): An Ethereum standard giving an agent an on-chain identity: an ERC-721 registry entry pointing at its metadata, plus reputation and validation registries.; See it on Sato Hub →Source ↗
Evidence tier: The label on every agent-economy number saying how it was obtained: 1 proof (read from the contract), 2 indicator (sampled or joined), 3 inference (a proxy).; See it on Sato Hub →Source ↗
Fee tier: The swap fee a concentrated-liquidity pool charges, fixed per pool — so one pair has several separate pools, each with its own liquidity and volume.; See it on Sato Hub →Source ↗
Graduation: The moment a bonding-curve token accumulates enough reserve to move to a conventional liquidity pool. Far fewer tokens graduate than launch.; See it on Sato Hub →Source ↗
Holder concentration: How much of a token's supply sits in how few addresses. Sato Hub reports it as not checked rather than guessing it from a partial holder list.; See it on Sato Hub →Source ↗
Impermanent loss: The shortfall a liquidity provider takes versus simply holding the two tokens, caused by the pool rebalancing against them as the price moves.; See it on Sato Hub →Source ↗
Liveness: Sato Hub's recency label — Active, Recent, Quiet or Dormant — from a listing's latest commit or release, never a post. A maintenance signal, not a risk grade.; See it on Sato Hub →Source ↗
MCP server: A process that exposes tools, resources and prompts to an AI agent over the Model Context Protocol, so the agent discovers what it can call at connect time.; See it on Sato Hub →Source ↗
MCP transport (stdio and Streamable HTTP): How an MCP client reaches a server: stdio launches it as a local subprocess; Streamable HTTP posts JSON-RPC to a URL. The tools are identical either way.; See it on Sato Hub →Source ↗
Non-custodial routing: Recommending where a swap, payment or job should go without ever holding, forwarding or signing for the funds involved.; See it on Sato Hub →Source ↗
Observed success (share of our checks): The percentage of Sato Hub's own daily checks of a hosted endpoint that succeeded over a window. It is a record of our checks, never the server's uptime.; See it on Sato Hub →Source ↗
Olas service: On Olas (Autonolas), a registered composition of agent instances run by bonded operators — the unit that actually runs, as opposed to a single agent.; See it on Sato Hub →Source ↗
Paymaster: A contract that pays gas for someone else's user operation, so an account can transact without holding the chain's native token.; See it on Sato Hub →Source ↗
Preflight verdict: One of four answers — go, caution, no, unknown — returned before an agent installs, connects, pays or trades, with every reading and the field it came from.; See it on Sato Hub →Source ↗
Reward recipient split: The on-chain configuration setting who receives a launched token's trading fees, and in what share — usually creator, platform and any referring interface.; See it on Sato Hub →Source ↗
SATI (8004-solana): A Solana program that ports the ERC-8004 agent-registry model: agent identities as on-chain accounts, readable by anyone without the operator's cooperation.; See it on Sato Hub →Source ↗
Sato Score: A 0–100 measure of how open, active and verifiable a listed project is, computed daily from evidence. Not a safety, quality or returns grade.; See it on Sato Hub →Source ↗
Session key: A short-lived key granted narrow, expiring permission to act for an account — so an agent can transact without ever holding the account's main key.; See it on Sato Hub →Source ↗
Skill disclosure: A static scan of an agent skill's own files, listing what it contacts, whether it generates keys or asks for credentials, and what tools it grants.; See it on Sato Hub →Source ↗
Slippage: The gap between a quoted price and the price actually executed, caused by pool depth, the size of the trade, and whatever happens between quote and fill.; See it on Sato Hub →Source ↗
Stablecoin settlement: Paying in a token pegged to a currency rather than a volatile asset — what makes a machine-to-machine price quotable and an agent's balance predictable.; See it on Sato Hub →Source ↗
Token launchpad: A contract system that deploys a token, seeds its initial market and configures fee recipients in one transaction — the entry venue for most agent tokens.; See it on Sato Hub →Source ↗
x402: An HTTP payment protocol: a server answers 402 with a price, the client pays in stablecoins, and retries the request with proof of payment attached.; See it on Sato Hub →Source ↗
x402 Bazaar: Coinbase's discovery index of x402-priced resources: sellers register an endpoint and price, and the catalogue is readable without a key.; See it on Sato Hub →Source ↗
x402 facilitator: A service that verifies an x402 payment payload and submits the settlement transaction on the server's behalf, so the server never touches a chain.; See it on Sato Hub →Source ↗

Reference

QuestionsWhat people ask about this pageShowHide

SourcesWhere the data comes fromShowHide

Cite this pagePlain text, APA or BibTeXShowHide

Cite this page

Sato Hub. "Onchain agent glossary." Sato Hub, updated 2026-10-24, accessed 2026-10-24. https://satohub.ai/glossary

Data last refreshed 2026-10-24; this page is rebuilt daily. Citations carry the date so a reader can tell which snapshot a claim came from. Catalog data is licensed CC-BY-4.0.

Related: the standards pages list what actually implements each of these, and the wording rules set out how this site words its claims.

Onchain agent glossary

Questions people ask about this glossary

Sources — the specifications these terms are defined by

Cite this page