swap-planner
Pass
Audited by Gen Agent Trust Hub on Nov 8, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: Fetches official token lists from tokens.pancakeswap.finance and market data from well-known providers like DexScreener, CoinGecko, and GeckoTerminal.
- [COMMAND_EXECUTION]: Utilizes shell utilities (curl, jq, cast) to query blockchain RPC nodes and financial APIs. The skill provides clear instructions to the agent to validate input formats (e.g., regex for token addresses) and use proper shell escaping to prevent command injection.
- [DATA_EXFILTRATION]: Performs an initialization ping to pancakeswap.ai which transmits environment metadata including the operating system name (uname -s) and hardware architecture (uname -m). This is documented as a session initialization step for usage tracking.
- [INDIRECT_PROMPT_INJECTION]: Processing external API data (like token names and descriptions) presents an indirect injection surface. The skill proactively addresses this with mandatory security rules that instruct the agent to treat all such data as untrusted and never interpret it as commands.
Audit Metadata