The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill implements a feature where [blocked_by] and [task_id] tags inside task body text are automatically parsed to update task metadata and dependency graphs. This represents an ingestion point for untrusted data that can influence the agent's understanding of task priority and order.
Ingestion points: Task bodies and response messages processed via yy ledger create and yy ledger mark commands.
Boundary markers: Absent. The skill does not define specific delimiters for separating user-provided body text from agent instructions.
Capability inventory: Shell command execution (yy ledger), filesystem writes via report paths, and local script execution.
Sanitization: Absent. The skill relies on the CLI tool's internal parsing of specific markup tags.
[COMMAND_EXECUTION]: The skill instructions utilize the yy ledger CLI tool to perform various task management operations. The skill explicitly enables shell directives to allow the agent to execute these commands. Several commands allow for specifying arbitrary file paths for reports and plans (e.g., --report, --plan-file, --target-bytes), which facilitates file system interaction and management of external data receipts.
[DYNAMIC_EXECUTION]: The skill documentation references the execution of a local Python script (.juno_task/scripts/controller_resolver.py) to diagnose controller registration. This involves executing code from a project-relative path at runtime.

kanban-workflow