dex-execution
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests external API data that determines transaction parameters, creating an injection surface.\n
- Ingestion points: Swap quote and token metadata retrieved from the Jupiter API in SKILL.md and scripts/get_quote.py.\n
- Boundary markers: Mandatory quote display and manual confirmation steps are required before execution (SKILL.md Step 2 & 3).\n
- Capability inventory: The skill performs cryptographic signing (solders) and network broadcasting (httpx) to Solana RPC nodes.\n
- Sanitization: Safety validation logic in references/safety_checklist.md filters for high price impact, excessive slippage, and mint mismatches.\n- [COMMAND_EXECUTION]: The skill provides Python scripts and code snippets for generating and submitting signed blockchain transactions.\n- [EXTERNAL_DOWNLOADS]: Functional dependencies include external Python libraries httpx and solders for API communication and blockchain operations.
Audit Metadata