tia-multiuser
Pass
Audited by Gen Agent Trust Hub on Aug 16, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill provides instructions for the agent to execute destructive lifecycle operations within the TIA project server environment, such as deleting project connections and local sessions.\n
- Evidence: Methods for DeleteConnection, DeleteLocalSessionFromServer, and CloseAndCommit (which discards pending changes) are exposed in SKILL.md and references/v21-api-surface.md.\n
- Mitigation: The instructions explicitly require remote-write authorization from the user before these operations are initiated.\n- [PROMPT_INJECTION]: The skill has an indirect prompt injection surface as it processes server names, project aliases, and project metadata retrieved from the environment.\n
- Ingestion points: Data entering the context via ProjectServer.GetServerProjects() and project aliases defined in workflows-and-safety.md.\n
- Boundary markers: The skill includes enforcement rules to resolve resources exactly and specifically forbids using non-deterministic methods like .First() or [0] to select objects.\n
- Capability inventory: High-privilege capabilities include deleting server connections, committing changes to the project server, and session deletion.\n
- Sanitization: Safety relies on mandatory authorization gates and exact identifier resolution rather than programmatic input sanitization.
Audit Metadata