opik-online-eval
Pass
Audited by Gen Agent Trust Hub on Sep 21, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from production traces to verify evaluation rules and map variables, creating an indirect injection surface.
- Ingestion points: Trace data is ingested via
client.search_traces()in Step 2 (to map variables) and Step 5 (to verify rule execution) withinSKILL.md. - Boundary markers: No explicit boundary markers or delimiters are defined in the instructions to prevent the agent from misinterpreting instructions embedded within trace content.
- Capability inventory: The skill utilizes
Bashto execute Python scripts that perform network operations (Opik REST API) and sensitive file reads (~/.opik.config). - Sanitization: There are no explicit instructions for sanitizing or escaping the content of production traces before the agent processes them.
Audit Metadata