langsmith-trace-analyzer
Pass
Audited by Gen Agent Trust Hub on Sep 7, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill's downloader scripts (
scripts/download_traces.pyandscripts/download_traces.ts) execute an external CLI toollangsmith-fetchusing safe subprocess methods (subprocess.runandspawnwith list-based arguments). This prevents shell injection vulnerabilities. - [EXTERNAL_DOWNLOADS]: The skill facilitates downloading trace data from LangSmith, a well-known service. It utilizes the
langsmithSDK and recommends the installation oflangsmith-fetchfor retrieving full trace payloads. - [INDIRECT_PROMPT_INJECTION]: The skill is an analysis tool for LangSmith traces, which contain external and untrusted model/user data. Ingestion points: Trace JSON files are loaded in
scripts/analyze_traces.py. Boundary markers: The generated reports do not use specific delimiters to isolate trace content. Capability inventory: The skill can execute local scripts and thelangsmith-fetchbinary. Sanitization: Trace data is processed for metrics but not sanitized for instruction-like content.
Audit Metadata