e2e-testing
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: Provides instructions for building and running the CLI from source using npm and node, as well as using tmux for interactive terminal UI testing. These are standard developer operations for local testing and debugging.- [DATA_EXFILTRATION]: Recommends the use of the --openai-logging flag which saves raw API request and response payloads to the /tmp/api-logs directory. While intended for debugging, this can lead to the exposure of sensitive prompts on multi-user systems.- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the inspection of raw API traffic logs. This creates a surface where malicious instructions embedded in model-generated data could be processed by the agent or developer. This is considered a low-risk inherent property of debugging model traffic.
Audit Metadata