langsmith-dataset

Pass

Audited by Gen Agent Trust Hub on Mar 4, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • Data Ingestion Considerations: The skill is designed to process external JSONL trace files to generate evaluation datasets. This involves a data ingestion surface where external content is incorporated into agent-managed data structures:
  • Ingestion points: The scripts generate_datasets.py and generate_datasets.ts read trace data from local file paths provided via the --input flag.
  • Boundary markers: The extracted content is processed without the addition of specific delimiters or instructions intended to isolate embedded data from agent commands.
  • Capability inventory: The skill is capable of writing to the local file system (JSON/CSV) and interacting with the LangSmith API for data uploads.
  • Sanitization: Trace fields are extracted and converted to strings or JSON representations without applying additional sanitization or filtering logic.
  • Credential Handling: The skill securely manages LangSmith API keys using environment variables, adhering to recommended security practices for service authentication.
  • Vendor Infrastructure: Network requests are directed exclusively to the official LangSmith service (api.smith.langchain.com), which is the intended destination for the vendor's dataset tools.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 4, 2026, 12:33 AM