tx-explain
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external blockchain sources which could contain malicious content, but it implements proper defensive measures.
- Ingestion points: External transaction data, receipts, and contract ABIs are fetched from Base RPC and Etherscan APIs (SKILL.md).
- Boundary markers: The instructions include a specific warning to treat all decoded data, symbols, and addresses as untrusted.
- Capability inventory: Uses standard network and parsing tools (curl, jq) and local messaging utilities.
- Sanitization: The skill mandates that untrusted data should never be interpolated beyond specific quoted variables, preventing command or prompt leakage.
- [EXTERNAL_DOWNLOADS]: The skill fetches blockchain-related data from well-known infrastructure providers including the Base mainnet RPC and Etherscan's API service. These references are for legitimate operational purposes and target established services.
- [COMMAND_EXECUTION]: Shell commands are used for making API requests and processing JSON data. Security is maintained by using a specialized wrapper for sensitive API keys, ensuring credentials are never exposed in the process environment or command history.
Audit Metadata