langfuse
Pass
Audited by Gen Agent Trust Hub on Apr 11, 2026
Risk Level: SAFE
Full Analysis
- [DATA_EXPOSURE]: The skill is designed to transmit LLM interaction data, such as prompts and responses, to the Langfuse backend for observability and evaluation purposes. This is the intended functionality of the documented tool and follows official integration patterns for a well-known service.
- [PROMPT_INJECTION]: This skill documents the handling of untrusted user data by logging it into the Langfuse ecosystem for monitoring. Ingestion points: User-provided message content passed to initialization and generation logging methods in SKILL.md. Boundary markers: None; the skill is intended to log the full context of the model interactions. Capability inventory: Network communication to the Langfuse service (cloud.langfuse.com) to store and visualize traces. Sanitization: Not present in the integration snippets as the primary goal is to record raw interaction data for analysis.
Audit Metadata