The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests task titles and bodies from the agkan backlog and interpolates them into a sub-agent's prompt for review.
Ingestion points: Task data retrieved via agkan task list --status backlog --json and agkan task get <id> --json in SKILL.md.
Boundary markers: The prompt uses Markdown headers (## Task Information) to separate data, but lacks explicit instructions for the sub-agent to ignore or sanitize instructions potentially embedded within the task title or body.
Capability inventory: The sub-agent is instructed to update task statuses, decompose tasks into subtasks, and research the codebase using available tools.
Sanitization: No evidence of sanitization or escaping of the task content before it is passed to the sub-agent.
[COMMAND_EXECUTION]: The skill executes local shell commands using the agkan CLI and jq for configuration management and task retrieval.
Evidence: agkan config get, agkan task list, and agkan task get commands are used to drive the workflow.

agkan-planning