The Agent Skills Directory

[COMMAND_EXECUTION]: The skill implements a sophisticated orchestration system using a local Node.js runtime (located in references/scripts/). It executes CLI commands to manage run identifiers, checkpoints, and state transitions, which is a standard pattern for L2 Domain Coordinators.
[COMMAND_EXECUTION]: A multi-agent advisor pattern is facilitated by references/agents/agent_runner.mjs. This script allows the coordinator to spawn external CLI agents (such as Claude or Codex) as subprocesses to perform architectural and traceability validation of task plans.
[PROMPT_INJECTION]: The skill exposes an indirect prompt injection surface in Phase 3 (Readiness Gate), where user-provided story data and acceptance criteria are interpolated into a prompt template (traceability_validator.md) for the advisor agent. This is a low-risk architectural feature, as the coordinator includes mandatory readiness scoring and human-in-the-loop checkpoints before any plan is implemented.
[SAFE]: The skill uses the Skill tool for delegation to worker skills (ln-301-task-creator and ln-302-task-replanner), following established orchestration patterns for hierarchical agent systems.

ln-300-task-coordinator