hermes-worker-provisioning
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill facilitates the management of Hermes worker profiles through local commands such as
hermes profile useandhermes tools list. These operations are consistent with the skill's provisioning purpose and do not invoke suspicious third-party binaries or network listeners. - [INDIRECT_PROMPT_INJECTION]: The skill manages the attack surface associated with processing task descriptions. Ingestion points: Task descriptions used to determine worker roles (found in SKILL.md). Boundary markers: None explicitly mentioned. Capability inventory: Execution of
hermesCLI for profile lifecycle management. Sanitization: The skill mitigates injection by instructing the agent to map natural language task text to a canonical list of approved roles (e.g.,reviewer,data-analyst) instead of accepting arbitrary names. - [DATA_EXPOSURE]: While the skill involves commands that display configuration (
hermes config show), it explicitly advises in step 4 to 'never copy secrets or broad session state by default,' which serves as a proactive measure against accidental credential exfiltration.
Audit Metadata