langsmith-fetch
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
1. Command Execution\n- Searching for sensitive files: The skill uses a find command to locate .env files in the directory tree.\n
- File:
SKILL.md\n - Evidence:
find . -name \".env\" -type f 2>/dev/null | head -5\n - Context: This is used to find the
LANGSMITH_API_KEYrequired for the skill to function. Searching for environment files is a sensitive operation that should be monitored.\n\n### 2. External Downloads\n- Dependency installation: The skill relies on thelangsmith-fetchpackage.\n - File:
SKILL.md\n - Evidence:
uv run --env-file <path> langsmith-fetch traces\n - Context: The tool uses
uvto run thelangsmith-fetchpackage, which may involve downloading it from a package registry if not already present. As this is a standard developer tool from the vendor, this is a routine operation.\n\n### 3. Indirect Prompt Injection\n- Ingestion of untrusted trace data: The skill fetches conversation traces and tool outputs for analysis.\n - File:
SKILL.md\n - Evidence:
langsmith-fetch traces ./debug --limit 10\n - Context: Traces represent historical data that can include content from external users or systems. This creates a surface for indirect prompt injection where malicious instructions stored in a trace could affect the agent during the debugging process.
Audit Metadata