collect-fees
Pass
Audited by Gen Agent Trust Hub on Nov 8, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses shell commands to manage a temporary filesystem, install Node.js dependencies, and execute helper scripts. It properly employs single-quoting for user-provided variables to prevent shell injection.
- [EXTERNAL_DOWNLOADS]: Fetches several official PancakeSwap and web3 utility packages from the NPM registry. It also retrieves token metadata and price data from public APIs including DexScreener and PancakeSwap Explorer.
- [DATA_EXFILTRATION]: Includes a telemetry ping to the vendor's domain (
pancakeswap.ai) during initialization. This sends non-sensitive metadata such as the agent identifier, timestamp, and host operating system/architecture for usage analytics. - [INDIRECT_PROMPT_INJECTION]: The skill processes external data (token names, symbols, and API responses) that could theoretically contain malicious instructions. However, the skill provides explicit security rules instructing the agent to treat this data as untrusted and display it verbatim without interpretation.
Audit Metadata