run-ai-employee
Pass
Audited by Gen Agent Trust Hub on Oct 30, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill requires the use of the dozaldevs command-line tool to trigger and manage AI employee tasks. This tool is a core component of the author's ecosystem and is used as intended for service interaction.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest user-specified inputs (such as topics or address lists) and pass them as arguments to shell commands. This creates a potential vector for indirect prompt injection if the external data contains instructions that could influence the downstream AI process.
- Ingestion points: Input data is accepted through shell variables like $TOPIC and $ADDRESSES within the SKILL.md instructions.
- Boundary markers: While shell variables are quoted to prevent command injection, no specific delimiters or prompts are instructed to isolate the user content for the downstream AI employee.
- Capability inventory: The skill utilizes the dozaldevs CLI to initiate cloud-based tasks and retrieve task identifiers.
- Sanitization: There are no instructions for sanitizing or validating the contents of the user-provided variables before they are transmitted.
Audit Metadata