badwhale-api-skill

Pass

Audited by Gen Agent Trust Hub on Jun 5, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: No security vulnerabilities were identified in the analyzed files.\n- [DATA_EXPOSURE]: The skill handles API credentials through environment variables (BADWHALE_API_KEY and BADWHALE_BASE_URL) rather than hardcoding. The execution script bw.py includes a _sanitize function that redacts authorization headers and API keys from its output, ensuring that sensitive tokens are not stored in the agent's conversation history or logs.\n- [COMMAND_EXECUTION]: API interactions are managed by a local Python script (skill/scripts/bw.py) using standard libraries. The skill requires an explicit confirmation step before submitting any transaction to the API, preventing accidental or unauthorized executions.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes external data from the BadWhale API to display supported chains and test results. The risk is minimized by the following evidence chain:\n
  • Ingestion points: API responses from /chains and /batches/{batchId} handled in skill/scripts/bw.py.\n
  • Boundary markers: Data is processed as structured JSON before being rendered to the user.\n
  • Capability inventory: Shell command execution of the local Python wrapper.\n
  • Sanitization: The _sanitize function in bw.py specifically removes potential injection or credential strings from the output provided to the agent.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 5, 2026, 08:49 AM
Security Audit — agent-trust-hub — badwhale-api-skill