execution-mode
Pass
Audited by Gen Agent Trust Hub on May 6, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill defines a mode that instructs the agent to bypass standard safety confirmation steps, explicitly stating "Do not pause for confirmation between tasks" and "Execute the ENTIRE TODO list without stopping." This behavior removes user-in-the-loop oversight for subsequent actions.
- [COMMAND_EXECUTION]: The instructions encourage the agent to execute complex task lists autonomously and use multiple tasks concurrently. This increases the potential impact of unverified task execution, especially if the TODO lists are derived from untrusted external data.
- [PROMPT_INJECTION]: Indirect Prompt Injection Surface Analysis:
- Ingestion points: The skill operates on TODO lists and task instructions which may originate from user input or project files (SKILL.md).
- Boundary markers: Absent; there are no instructions to delimit or ignore embedded instructions within the tasks being executed.
- Capability inventory: The execution mode leverages the agent's ability to execute shell commands, modify files, and manage subtasks (SKILL.md).
- Sanitization: Absent; the skill does not include logic to validate or sanitize the tasks before autonomous execution.
Audit Metadata