fep-agent-hub
Pass
Audited by Gen Agent Trust Hub on Nov 3, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill facilitates the execution of local repository scripts for regression testing and case validation (e.g.,
scripts/mcp_heat_smoke.py). It explicitly prohibits the use of arbitrary shell commands or macros, requiring the use of fixed runners and structured parameters to maintain an auditable environment. - [INDIRECT_PROMPT_INJECTION]: The skill processes external data such as STEP geometry files and manifests. It mitigates potential injection or data corruption risks by implementing a mandatory 'Universal acceptance stack' described in
references/validation-and-claims.md, which requires each stage of the workflow (geometry, mesh, solver input) to pass validation before the next stage begins. - [DYNAMIC_EXECUTION]: While the skill involves generating domain-specific Solver Input Files (SIF) for Elmer FEM, this behavior is constrained by allowlisted profiles. The orchestration workflow includes a specific
elmer_sif_validatestep to ensure that generated solver instructions are safe and conform to expected physics gates before the native solver is invoked.
Audit Metadata