running-chat-assistant
Pass
Audited by Gen Agent Trust Hub on Sep 28, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONCREDENTIALS_UNSAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill relies on the execution of the
robomotionCLI andgitcommands to manage automation flows, agent instances, and local robot connections. - [EXTERNAL_DOWNLOADS]: Recommends downloading the
robomotionCLI from the vendor's official website (robomotion.io/downloads) when an update is required. - [INDIRECT_PROMPT_INJECTION]: The
robomotion agent chatcommand accepts user-supplied text as arguments to test chatbot responses, creating a surface for indirect injection. - Ingestion points: The
chatcommand argument (SKILL.md). - Boundary markers: Instructions recommend using double quotes for chat inputs.
- Capability inventory: Execution of workspace management commands and network interaction via the vendor CLI tool.
- Sanitization: No explicit sanitization or filtering of the user-provided text is documented.
- [CREDENTIALS_UNSAFE]: Mentions the storage of session tokens in
.robomotion/agent.jsoncreated by authentication commands. The skill includes specific warnings for the agent to avoid leaking these tokens or session IDs in outputs.
Audit Metadata