execute-icebox-subtask
Pass
Audited by Gen Agent Trust Hub on Apr 8, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill uses the
agkanCLI tool to manage task workflows. The operations (get, update) are consistent with the stated purpose of reviewing and promoting icebox tasks. - [COMMAND_EXECUTION]: Executes shell commands using the
agkantool and standard utilities likecat. These are used for task retrieval and status updates. - [INDIRECT_PROMPT_INJECTION]: The skill processes external task data (via
agkan task get). While this represents a surface for indirect prompt injection if task content contains malicious instructions, the skill's capabilities are limited to task status updates within theagkansystem. - Ingestion points: Reads task content from the
agkandatabase (SKILL.md). - Boundary markers: Standard shell interpolation is used; instructions provide guidance on evaluation.
- Capability inventory: Task retrieval and status updates via the
agkanCLI (SKILL.md). - Sanitization: Content is written to a temporary file via heredoc to maintain integrity before update.
Audit Metadata