execute-task
Pass
Audited by Gen Agent Trust Hub on Apr 8, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes various shell commands including 'git pull', 'git checkout', and several 'agkan' CLI operations to manage task states and repository data.
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection. It retrieves task titles and bodies from the external 'agkan' tool and interpolates them directly into the instructions for a spawned sub-agent.
- Ingestion points: Data enters the context via 'agkan task list' and specific task detail fetches.
- Boundary markers: None are used to separate untrusted task content (Title/Body) from the agent's instructions in the Task tool prompt.
- Capability inventory: The skill and its sub-agents can execute shell commands (git, agkan), read local files (SKILL.md), and spawn further sub-agents.
- Sanitization: There is no evidence of validation or escaping for the 'title' or 'body' fields before they are passed to the sub-agent.
Audit Metadata