The Agent Skills Directory

[SAFE]: The skill implements a robust multi-agent orchestration framework with clearly defined roles (Planner, Worker, Explorer, Validator, Reporter) and strict operational boundaries that ensure task segregation.
[SAFE]: No prompt injection or instruction override patterns were detected; the guidelines focus on operational efficiency, consistency, and autonomous task execution based on approved plans.
[SAFE]: External URLs cited in the documentation (e.g., Anthropic documentation and technical blogs like PubNub) are used exclusively for informational reference and do not involve remote code execution or data exfiltration.
[SAFE]: Terminal output and interaction rules are designed to maintain a clean interface for the user while performing autonomous background tasks.
[SAFE]: Capability usage, including shell command execution for builds and web navigation for research, is strictly aligned with the primary purpose of a software development orchestrator.
[SAFE]: The orchestration logic enforces the review and approval of plans before execution, providing a human-in-the-loop safety checkpoint for autonomous actions.

workflow-agent