execute-subtask
Pass
Audited by Gen Agent Trust Hub on Apr 8, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: Indirect prompt injection surface identified. The skill retrieves task content using
agkan task get <id> --jsonand instructs the agent to implement logic based on that task content, creating a pathway for external instructions to influence agent actions.\n - Ingestion points:
SKILL.md(viaagkan task get).\n - Boundary markers: Absent; task content is used directly to guide implementation.\n
- Capability inventory:
git(branching and pushing),gh(PR creation), andagkan(task status and body updates) subprocess calls inSKILL.md.\n - Sanitization: Absent; the skill does not validate or filter task body content before use.\n- [COMMAND_EXECUTION]: The skill executes multiple shell commands to manage the development lifecycle, including
git checkout,git commit,git push, andgh pr create. It also uses theagkanCLI tool for task management and performs file writes to the/tmpdirectory using here-docs.
Audit Metadata