intercom
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses
subprocess.Popeninscripts/agy_ipc.pyto manage a background IPC hub daemon. This is used safely to execute the script itself using the current Python interpreter (sys.executable) to establish the communication mesh. - [COMMAND_EXECUTION]: The
scripts/test_ipc.pyfile usessubprocess.runto execute integration tests. This is a standard practice for verifying the functionality of the IPC transport and protocol. - [INDIRECT_PROMPT_INJECTION]: The skill ingests data from other local agent sessions. It implements several defense-in-depth measures to mitigate risks, including the use of a dedicated Communications Officer subagent to isolate network traffic, enforcing restricted file permissions (0700) on IPC directories, and providing an inbox reset mechanism to ensure stale messages are not replayed into the agent's context.
Audit Metadata