task-checkout
Pass
Audited by Gen Agent Trust Hub on Aug 15, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes task details which acts as a vulnerability surface for indirect instructions.
- Ingestion points: The agent ingests data from
~/.ai-software-agency/tasks.mdand CEO delegation messages. - Boundary markers: The skill lacks explicit delimiters or instructions to ignore embedded commands within task titles or metadata.
- Capability inventory: The agent possesses capabilities to write to local registry/log files and invoke the
TaskUpdatetool. - Sanitization: There is no evidence of sanitization or validation of the
{id},{agent-name}, or task metadata before they are interpolated into tool calls and file operations. - [DATA_EXPOSURE_AND_EXFILTRATION]: The skill accesses and updates files in
~/.ai-software-agency/. This behavior is consistent with the skill's primary function of maintaining a shared task coordination registry.
Audit Metadata