langfuse
Pass
Audited by Gen Agent Trust Hub on Nov 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill establishes an attack surface for indirect prompt injection by facilitating the ingestion and processing of untrusted external data (LLM inputs/outputs) into an observability pipeline.
- Ingestion points: Data enters through instrumentation points defined in
SKILL.md, specifically via the Langfuse SDK's observation and prompt management features. - Boundary markers: The skill instructions recommend the use of masking hooks and allowlists to control data boundaries, though these do not prevent injection within the ingested content itself.
- Capability inventory: The skill utilizes the
langfuseSDK for external data transmission; no arbitrary code execution or unauthorized file system access capabilities are present in the skill's logic. - Sanitization: The procedure mandates the use of
mask_otel_spansand synthetic requests to filter or verify content before it is exported to the external collector.
Audit Metadata