"Build a trading agent" sounds like one thing to build. It isn't. A trading agent needs a strategy layer that decides what to do, and an execution layer that can actually move funds on a chain — and those two layers are built by different projects, for different reasons. Sato Hub's library tracks 374 resources across the onchain agent stack; here's the honest state of the eight that map onto trading specifically, layer by layer, sourced to each project's Sato Hub page.
The strategy layer: deciding what to trade
[TradingAgents](https://satohub.ai/resources/tradingagents) is the most direct hit: an open-source multi-agent LLM framework that simulates a trading firm, with separate analyst, trader, and risk-management agents debating a position before it's taken. It's explicitly built for research — a simulation of how a trading desk reasons, not a production execution engine. That distinction matters more than it sounds: a framework that models risk-management as a *role* inside a debate is not the same as a framework that enforces risk limits on a real wallet.
[ElizaOS](https://satohub.ai/resources/elizaos) and [Olas (Autonolas)](https://satohub.ai/resources/olas) sit one level up, as general-purpose agent frameworks that trading logic gets built on top of rather than shipped with. ElizaOS is an open-source TypeScript framework for crypto-native multi-agent systems, active across Solana, Ethereum, and Base. Olas is a network and framework for co-owned autonomous services running onchain, spanning Ethereum, Gnosis, and Base. Neither is a trading bot out of the box — both are the scaffolding a builder wires a strategy into.
[Almanak](https://satohub.ai/resources/almanak) is the one project here purpose-built for the DeFi trading use case specifically: an agent framework paired with a non-custodial vault protocol for designing, backtesting, and deploying automated yield and trading strategies, with reach across eleven chains including Base, Arbitrum, and Solana. Backtesting-as-a-feature is rare enough in this list to call out on its own.
The execution layer: agents that can actually move funds
A strategy is only as good as the wallet underneath it. [Coinbase AgentKit](https://satohub.ai/resources/coinbase-agentkit) is the developer-tool entry: Coinbase's toolkit for giving agents wallets and the ability to take onchain actions on Base, Ethereum, and beyond. [GOAT SDK](https://satohub.ai/resources/goat-sdk), from Crossmint, is a skill repo of onchain actions agents can call across Ethereum, Base, and Solana — the plumbing that turns "swap this token" from an idea into a signed transaction. [Solana Agent Kit](https://satohub.ai/resources/solana-agent-kit), built by SendAI, does the same job scoped to Solana protocol actions specifically.
[0xGasless AgentKit](https://satohub.ai/resources/0xgasless-agentkit) covers a narrower but real gap: gasless wallet access for agents on BNB Chain, Avalanche, and Base, so an agent doesn't need to hold native gas tokens on every chain it trades across.
What's still missing
Line the eight up and the gap is visible: strategy and execution are built by different teams, on different chains, with no shared standard connecting a "decide to trade" agent to a "sign this transaction" wallet. None of these eight is a single project that reasons about a trade *and* owns the risk controls on the wallet that executes it — that composition is still on the builder to wire together. And every trust score above comes from Sato Hub's own Sato Score, which measures how open, active, and verifiable a project is — not its trading performance or safety, and it's not evidence any of these strategies works.
What to watch
The strategy side is crowding — TradingAgents, ElizaOS, Olas, and Almanak all approach the same problem from different angles, with Almanak alone owning backtesting as a named feature. The execution side is thinner and more chain-fragmented: three chain-specific wallet toolkits and one gasless layer, no single execution SDK spanning all of them. Whoever builds the connective layer — a shared way for a strategy agent to hand its decision to a wallet agent with enforced limits — closes the actual gap.
Want to build one? Start with the stack, not a blank prompt: [satohub.ai/build](https://satohub.ai/build).