hermes-agent-mission-control
Pass
Audited by Gen Agent Trust Hub on Sep 12, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSPERSISTENCECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructs the user to install and run the 'minionsai' package from the NPM registry using 'npm' or 'npx'. This package is a primary resource provided by the skill's authoring vendor for managing autonomous agent workflows.\n- [PERSISTENCE]: The skill includes a 'Routines' feature designed to schedule and track recurring autonomous agent tasks, which functions as a management mechanism for persistent agent activities.\n- [COMMAND_EXECUTION]: The troubleshooting section suggests shell commands for the user to execute manually, such as 'lsof' and 'kill', to identify and manage the Minions server process to resolve port conflicts.\n- [INDIRECT_PROMPT_INJECTION]: The mission control interface provides a surface for indirect prompt injection as it processes task descriptions and agent tool-call data, which are subsequently evaluated by a 'completion judge' LLM.\n
- Ingestion points: Task creation chat and live tool call streaming described in SKILL.md.\n
- Boundary markers: None identified in the provided documentation.\n
- Capability inventory: Local file system access (via file browser and SQLite storage) and autonomous tool execution (via Hermes sessions).\n
- Sanitization: No specific sanitization or validation mechanisms are detailed in the architecture overview.
Audit Metadata