langsmith-experiment-diagnosis
Pass
Audited by Gen Agent Trust Hub on Oct 27, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- Command Execution: The skill utilizes the
langsmithCLI to query experiment data and traces. It includes specific guidance to use quoted command arguments and avoid splicing untrusted content into shell commands, which helps mitigate command injection risks. - Indirect Prompt Injection Surface: As the skill processes dataset inputs, model outputs, and feedback from external sources, it has an inherent surface for indirect prompt injection. However, the instructions explicitly command the agent to treat this content as evidence and not as instructions, which is a recommended safety practice.
- Credential Handling: The skill maintains a security-first approach by instructing the agent not to ask for, display, or store API keys, relying instead on the user's existing authenticated CLI session.
Audit Metadata