monitor-and-review-ai-employee-work
Pass
Audited by Gen Agent Trust Hub on Oct 30, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes the
dozaldevsCLI to perform operations such as fetching task details, deciding on tasks, and listing execution events. This is the intended behavior for interacting with the vendor's platform.\n- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from task outputs and execution logs, which presents a surface for indirect prompt injection attacks.\n - Ingestion points: External content is brought into the agent context via
dozaldevs get-task(readingoutput.draft),dozaldevs get-task-events, anddozaldevs list-executions(reading transcripts).\n - Boundary markers: The instructions do not specify the use of delimiters or warnings to ignore instructions embedded within the task content or logs.\n
- Capability inventory: The skill has the capability to modify state via
dozaldevs decide-task(approve/reject).\n - Sanitization: There is no evidence of sanitization or filtering of the content fetched from the CLI before it is processed by the agent.
Audit Metadata