managing-experiment-lifecycle

Pass

Audited by Gen Agent Trust Hub on Sep 24, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill identifies a surface for indirect prompt injection where untrusted data can influence agent actions.
  • Ingestion points: The agent ingests experiment metadata (such as names, descriptions, and variant keys) and user-provided conclusion comments to execute lifecycle transitions (referenced throughout SKILL.md).
  • Boundary markers: The instructions do not specify the use of delimiters or provide explicit warnings to the agent to ignore instructions embedded within the data being processed.
  • Capability inventory: The skill utilizes tools that perform significant write operations, including modifying experiment states and triggering automated source code removal and pull request creation in connected GitHub repositories (experiment-ship-variant and experiment-end tools).
  • Sanitization: There is no evidence of validation, filtering, or escaping of external content before it is interpolated into prompts or used in operations.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 24, 2026, 01:10 AM
Security Audit — agent-trust-hub — managing-experiment-lifecycle