operate-ai-employees
Pass
Audited by Gen Agent Trust Hub on Oct 30, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The instructions direct the user to install the 'dozaldevs' package globally using 'npm install -g dozaldevs' or to execute it via 'npx -y dozaldevs'. These are legitimate vendor resources provided by the skill author 'dozaldevs'.
- [COMMAND_EXECUTION]: The skill operates by executing the 'dozaldevs' CLI tool for various account operations, including 'list-dozals', 'get-task', 'whoami', and 'run-app-action'.
- [INDIRECT_PROMPT_INJECTION]: The skill processes structured data from the DozalDevs API and natural language from the user, creating a surface for potential indirect prompt injection.
- Ingestion points: Ingests plain-language commands from users and JSON payloads returned by the DozalDevs production gateway (SKILL.md).
- Boundary markers: No specific delimiters or boundary warnings for untrusted data are provided in the skill instructions.
- Capability inventory: The agent can execute CLI commands that perform network requests and trigger actions in connected third-party applications (references/customer-api.md).
- Sanitization: The documentation does not describe specific sanitization or filtering logic applied to external content before it is interpreted by the agent.
Audit Metadata