agents-status
Pass
Audited by Gen Agent Trust Hub on Aug 3, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill interacts with the ai-coord CLI tool to perform session inventory and communication between agents. The commands are predefined and limited to coordination tasks.\n- [PROMPT_INJECTION]: The skill manages a surface for indirect prompt injection by ingesting session labels and messages from other agents. It includes an explicit evidence chain for mitigation: ingestion points are status reports and messages; boundary markers are present in the form of instructions; capability inventory is limited to coordination commands; and sanitization is enforced by directing the agent to treat external text as data only.
Audit Metadata