SRSWTI drops Shadows, distributed task framework

Shadows is a direct response to the "unreliable" state of current AI agent scheduling, prioritizing deterministic systems over the "ai slop" rush.

• –Uses Redis Streams for reliability and at-least-once delivery, a significant step up from basic cron or heartbeat loops
• –"Striking" allows pausing tasks by argument (e.g., team_id) without a full restart, essential for production stability
• –Optimized for mixed hardware environments, routing heavy inference to high-spec nodes while keeping light tasks local
• –Same-machine task routing drops overhead from ~2500µs to ~5µs, enabling 20,000 tasks per second
• –MIT licensed and built for Python 3.12+, it positions itself as the "engineered" layer for autonomous agent infrastructure