uniswap/uniswap-ai
7 skills4.9k installs1.5k starsGitHub
Install
npx skills add https://github.com/uniswap/uniswap-aiSkills in this repo
1Swap IntegrationSwap integration is an agent skill from Uniswap’s AI skill set that teaches advanced patterns for the Uniswap Trading API beyond a minimal quote-and-send swap. Solo builders shipping DeFi features, trading bots, or wallet-adjacent products can use it when they need executable calldata, correct permit handling, and optional ERC-4337 smart-account execution through delegations and a bundler. The documentation walks architecture from gateway quotes to on-chain targets, with TypeScript and viem on chains like Base, and calls out operational concerns such as rate limits and L2 WETH edge cases. It assumes you already understand basic swap UX and now must harden server-side or agent-driven execution for real users. This is integration knowledge, not a hosted MCP swap button: you still manage API keys, keys, and chain config. Use it while building backend swap orchestration before you ship to mainnet or scale automated rebalancers.1.1kinstalls2Viem Integrationviem-integration is a technical reference skill for accounts and keys in the viem TypeScript Ethereum library. Solo builders shipping on-chain features, bots, or DeFi integrations use it when they need to create accounts from private keys or mnemonics, attach them to createWalletClient, and understand safe key-handling patterns. The content centers on copy-pasteable snippets: privateKeyToAccount, mnemonicToAccount, generatePrivateKey, and mainnet wallet client configuration with http transport. It emphasizes never using well-known test keys or example seed phrases with real funds. This is phase-specific integration knowledge rather than a workflow spanning the whole product journey; pair it with your RPC provider, contract ABIs, and deployment skills. Agents should treat secrets as environment variables and avoid logging key material.675installs3Liquidity Plannerliquidity-planner is a data-provider reference skill for solo builders and agent operators who help users evaluate Uniswap liquidity positions using public market APIs. It centers on DexScreener for discovering pairs, reading USD liquidity and volume, and inspecting token prices across Layer-2 and mainnet deployments, with DefiLlama noted for supplemental APY data. The content is procedural: copy-ready HTTP calls and jq filters rather than a full risk model or position-sizing framework. Use it when your agent must ground LP recommendations in live pool stats instead of guesswork, especially on Base and other networks called out in the examples. You should still validate contracts, impermanent loss, and wallet security outside this skill. Complexity assumes comfort with REST APIs, shell, and DeFi vocabulary; it does not replace financial advice or audited smart-contract interaction skills.666installs4Swap PlannerSwap-planner is an agent skill from the Uniswap AI tooling line that teaches your coding agent how to fetch real-time token prices, pool liquidity, and pair discovery data when planning swaps on Uniswap-style markets. It centers on DexScreener as the primary HTTP API and DefiLlama as a fallback for prices and pool TVL, with concrete curl and jq examples for major EVM networks including Ethereum, Base, and Arbitrum. Solo and indie builders shipping DeFi copilots, portfolio agents, or internal treasury scripts use it to ground swap recommendations in observable liquidity instead of guessing from static token lists. The reference notes where coverage is thin so agents do not hallucinate deep books on minor chains. It is procedural knowledge for integration work—not a wallet executor—so you still implement signing, slippage guards, and compliance yourself. Pair it with your own execution and risk checks before any mainnet trade.662installs5V4 Security Foundationsv4-security-foundations is an agent skill that walks solo and indie builders through a Uniswap v4 hook pre-deployment audit using a structured markdown checklist. It targets developers shipping custom hooks on v4 who need more than informal code review before mainnet—covering who may call hook callbacks, whether returned deltas match real token movements, and whether settlement cannot leave the PoolManager in an inconsistent state. The skill is framed as procedural knowledge your coding agent applies while you read Solidity alongside it: verify msg.sender against the pool manager, validate admin paths cannot brick the contract, and trace take/settle/take flows for exotic tokens. Use it in the Ship phase when the hook compiles and tests pass but economic and permission bugs still dominate real-world exploit stories. It complements generic security linters by encoding Uniswap-specific invariants from the v4 hook lifecycle.631installs6DeployerDeployer is an agent skill for solo and indie builders shipping token auctions on EVM chains using Uniswap’s Continuous Clearing Auction factory. It walks you through a mandatory educational disclaimer, loading or collecting auction configuration, validating inputs against the target network’s factory address, surfacing a deployment plan, and only then issuing Foundry forge and cast commands. CREATE2 is emphasized so the same configuration yields predictable contract addresses when you repeat deploys on other chains. The skill is opinionated about safety: no deployment without explicit user confirmation after validation. It fits builders who already understand CCA parameters and need a repeatable, auditable deploy ritual inside Claude Code rather than ad-hoc shell one-liners. Post-deployment guidance closes the loop so you can verify what landed on-chain.620installs7ConfiguratorConfigurator is an agent skill for Uniswap-style Continuous Clearing Auction (CCA) smart contracts. It guides solo builders and small teams through an interactive bulk form: batching up to four questions per AskUserQuestion call, accepting custom addresses and numbers via Other or skip paths, and never blocking progress when a value is not ready yet. The flow covers auction timing, token metadata hooks, pricing mechanics, and supply schedule generation through dedicated cca-supply-schedule skills for generate and encode steps. It is aimed at founders preparing a primary sale or treasury distribution who need contract parameters encoded correctly rather than guessed in a generic chat. Model hint in upstream metadata targets Opus-class reasoning for parameter tradeoffs. When your IDE lacks AskUserQuestion, the same fields are collected conversationally so Cursor or generic agents can still complete the checklist.613installs