agkan-icebox-subtask
Pass
Audited by Gen Agent Trust Hub on Sep 21, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes shell commands to interact with the agkan CLI tool and manage local file operations. It specifically uses cat to write task bodies to temporary files in /tmp using the shell process ID ($$) for naming, which are then read back by the agkan update command.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests external data from task descriptions that could contain malicious instructions designed to influence the agent's evaluation and decision-making process.
- Ingestion points: Task content is retrieved via the agkan task get command in SKILL.md.
- Boundary markers: The skill does not implement delimiters or explicit instructions to the agent to ignore or isolate instructions found within the task body.
- Capability inventory: The skill possesses the capability to modify task statuses (backlog, closed) and overwrite task bodies using the agkan task update command.
- Sanitization: There is no evidence of sanitization, validation, or filtering of the task content before it is processed or re-uploaded to the task management system.
Audit Metadata