adams-evidence-mbd
Pass
Audited by Gen Agent Trust Hub on Nov 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external data from simulation results (
adams_evidence_latestandvalidation.json) to verify the success of dynamics tasks. This creates an indirect prompt injection attack surface where content within those files could potentially influence the agent's reasoning or subsequent tool calls. - Ingestion points: External data enters the context via the reading of
adams_evidence_latestandvalidation.jsonfiles. - Boundary markers: The instructions do not specify the use of delimiters (e.g., XML tags or triple backticks) to separate ingested simulation data from system instructions.
- Capability inventory: The skill has access to tools for environment probing, project creation, and simulation execution (
adams_simulation_run), which could be targeted by an injection. - Sanitization: There is no evidence of data sanitization or filtering applied to the simulation outputs before they are processed by the agent.
Audit Metadata