Sprout MCP

Routes your LLM tasks through a three-tier content pipeline to cut costs. Start drafts with Haiku at $5/M tokens, fact-check with Sonnet at $15/M, and deep-verify with Opus at $75/M only when needed. Exposes 13 tools including submit_chunk for storing content with provenance tracking, get_review_queue to see what needs escalation, and mark_reviewed to promote work between tiers. Ships with routing rules for common tasks like summarization, code review, and theological analysis, but you can override via JSON config. Tracks token spend per model and stores everything in SQLite. Reach for this when you're burning through Opus tokens on work that cheaper models could handle for first passes.