opik-openclaw-observability

Pass

Audited by Gen Agent Trust Hub on Sep 12, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill facilitates the integration of Opik, an open-source observability platform by Comet, into the OpenClaw agent environment for monitoring and tracing.
  • [EXTERNAL_DOWNLOADS]: The skill instructs users to install the @opik/opik-openclaw plugin using the platform's native CLI and references the official Comet GitHub repository (comet-ml/opik-openclaw). These are well-known services.
  • [DATA_EXFILTRATION]: Agent trace data, including prompts and tool results, is exported to the Opik API at comet.com. This network communication is the primary purpose of the observability skill.
  • [CREDENTIALS_UNSAFE]: The documentation follows security best practices by recommending the use of environment variables (OPIK_API_KEY) for managing sensitive credentials.
  • [INDIRECT_PROMPT_INJECTION]:
  • Ingestion points: The skill captures LLM inputs, outputs, and tool arguments at runtime to generate traces.
  • Boundary markers: Not explicitly mentioned in the configuration, but inherent to the tracing middleware's operation.
  • Capability inventory: Network transmission to the configured Opik endpoint.
  • Sanitization: The skill provides a toolResultPersistSanitizeEnabled configuration option to handle potentially sensitive content in tool outputs.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 12, 2026, 06:49 PM
Security Audit — agent-trust-hub — opik-openclaw-observability