The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes external item lists (files, modules) to be fanned out to agents, which represents an indirect prompt injection surface.
Ingestion points: Untrusted data enters via the $work argument and discovery commands (like git diff or grep) in Phase 0.
Boundary markers: The instructions do not define explicit markers or warnings to delimit untrusted data during processing.
Capability inventory: The skill orchestrates tasks using the Workflow, Agent, and Bash tools.
Sanitization: The skill does not specify any sanitization or filtering of item content before it is passed to agents.
[COMMAND_EXECUTION]: The skill provides logic templates in references/fanout-patterns.md for the Workflow tool to orchestrate concurrent agent actions.
Evidence: Orchestration logic includes the use of pipeline() and parallel() functions.
Analysis: These templates guide the agent in generating and executing local coordination scripts for task management.

fan-out-work