analyzing-experiment-session-replays
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from session recording metadata and experiment configurations which could potentially contain malicious instructions intended to influence the agent's behavior.
- Ingestion points: The skill ingests data from the
execute-sqltool (querying system tables likesystem.experimentsandsystem.feature_flags) and thequery-session-recordings-listtool (SKILL.md). - Boundary markers: Absent. The instructions do not provide delimiters or specific prompts to the agent to ignore instructions that might be embedded within the retrieved data.
- Capability inventory: The agent is granted the ability to execute HogQL via
execute-sqland retrieve session details viaquery-session-recordings-list(SKILL.md). - Sanitization: Absent. No explicit sanitization or validation steps are outlined for the metadata retrieved from system tables or session recordings before it is processed by the agent.
Audit Metadata