agkan-planning-subtask
Pass
Audited by Gen Agent Trust Hub on Oct 27, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests external task data and interpolates it into prompts for sub-agent delegation, which can be exploited if the task data contains malicious instructions.
- Ingestion points: Task metadata (title, ID) and the task body are retrieved using the
agkan task getcommand in Step 1 of SKILL.md. - Boundary markers: The workflow uses markdown headings to separate data fields but lacks specific instructions to the sub-agent to treat the interpolated content as passive data.
- Capability inventory: The skill has the ability to add new tasks, update task bodies, change task statuses, and invoke sub-agent workflows via the
Delegatepseudocode. - Sanitization: Task content is not sanitized or escaped before being placed into the delegation prompt or appended to task bodies.
Audit Metadata