Uniswap Execute Swap

Execute a Uniswap token swap. Use when user wants to swap, trade, buy, or sell tokens. Handles quotes, safety checks, simulation, and execution autonomously. Supports V2, V3, V4, UniswapX, and cross-chain routing on all supported chains.

Security Scan

The skill claims to autonomously execute Uniswap trades but does not describe where signing credentials/wallet access come from or how confirmations are handled, which is a major mismatch for a component that performs irreversible financial transactions.

Purpose & Capability

The name and description (execute swaps on Uniswap) match the instructions (resolve tokens, safety checks, delegate execution). However, executing a swap necessarily requires wallet signing (private key / wallet access), approvals, or a custodial API — none of which are declared (no required env vars, config paths, or credentials). The skill therefore omits a core capability it needs to operate.

Instruction Scope

SKILL.md delegates execution to Task(subagent_type:trade-executor) and calls MCP tools (search_tokens, check_safety_status) which is reasonable, but the instructions do not specify how the agent obtains or stores wallet keys, how user confirmation is obtained before on-chain execution, or what exactly the trade-executor subagent is permitted to do. The skill also states it operates "autonomously" but gives no required confirmation step, creating scope creep from 'suggest a trade' to 'execute with signing'.

Install Mechanism

There is no install spec (instruction-only), so nothing is written to disk by the skill itself — low install risk. It does, however, depend on an external Agentic-Uniswap MCP server (GitHub repo referenced). That external dependency is not bundled or verified here; use of an unverified MCP server is an operational and trust risk.

Credentials

The skill declares no required environment variables or credentials, yet performing swaps requires access to signing credentials and possibly allowlists/spending-limit configuration. That discrepancy means required secrets or config may be requested at runtime by the trade-executor or MCP server but are not documented here — a proportionality/opacity issue.

Persistence & Privilege

always is false (good) but disable-model-invocation is false (normal), meaning the agent can invoke the skill autonomously. Combined with the ability to execute financial transactions and the missing detail on confirmation and credential handling, autonomous invocation increases risk: the agent could attempt to execute trades without clear, documented user approval flows.

What to consider before installing

This skill claims to execute on-chain swaps but does not say how it will sign transactions or obtain wallet access. Before installing or enabling it: 1) Ask the author how signing is performed — does the trade-executor require your private key, a local wallet connector (e.g., WalletConnect), or a custodial API token? 2) Require an explicit, documented confirmation step before submitting any transaction. 3) Verify and audit the referenced MCP server code (https://github.com/wpank/Agentic-Uniswap) and the trade-executor subagent — do not trust an unknown third-party server with signing or funds. 4) Test on a testnet with small amounts first. 5) If you cannot confirm where private keys remain (local vs remote) and how approvals are handled, do not enable autonomous execution.

Execute Swap

Execute a token swap on Uniswap with full safety validation.

Activation

Use this skill when the user says any of:

• "Swap X for Y"
• "Buy X with Y"
• "Sell X for Y"
• "Trade X for Y"
• "Exchange X to Y"
• "Convert X to Y"

Input Extraction

Extract these parameters from the user's message:

Parameter	Required	Default	Source
tokenIn	Yes	—	Token name/symbol/address
tokenOut	Yes	—	Token name/symbol/address
amount	Yes	—	Numeric value
chain	No	ethereum	Chain name or context
slippage	No	0.5%	Explicit percentage
routing	No	auto	"via V3", "use UniswapX", etc.

Workflow

1. Validate inputs: Resolve token symbols using search_tokens. Confirm chain is supported.

2. Pre-flight safety check: Call check_safety_status to verify:

   • Spending limits have room for this trade
   • Rate limits are not exhausted
   • Circuit breaker is not tripped

3. Delegate to trade-executor: Launch Task(subagent_type:trade-executor) with:

   • tokenIn, tokenOut, amount, chain
   • slippageTolerance (in bps)
   • routingPreference (auto/v2/v3/v4/uniswapx)

4. Report result to the user in a clear format:

Swap Executed Successfully

  Input:  500.00 USDC
  Output: 0.1538 WETH ($499.55)
  Price:  1 WETH = $3,248.04
  Impact: 0.01%
  Gas:    $0.42

  Tx: https://basescan.org/tx/0xABC...

  Safety: All 7 checks passed

Error Handling

Error	User Message	Suggested Action
SAFETY_SPENDING_LIMIT_EXCEEDED	"This swap would exceed your $X daily limit."	Reduce amount or wait
SAFETY_TOKEN_NOT_ALLOWED	"TOKEN is not on your allowlist."	Add to config
SAFETY_SIMULATION_FAILED	"Swap simulation failed: [reason]."	Check addresses, try smaller
INSUFFICIENT_LIQUIDITY	"Not enough liquidity at acceptable slippage."	Try smaller amount

MCP server dependency

This skill relies on Uniswap MCP tools for chain support lookup, token search, safety checks, and swap execution. When used in isolation (for example, from a skills catalog), ensure the Agentic Uniswap MCP server is running:

• Repo: Agentic-Uniswap MCP server
• Package: @agentic-uniswap/mcp-server