OpenSquilla is a Python microkernel runtime for AI agents that reduces token costs by 60-80% via smart model routing. It features a four-tier memory system with local embeddings and a secure sandbox. The tool supports multiple communication channels and integrates with various LLM providers.
Highlights
Achieves 60-80% token cost savings through smart routing.
Features a four-tier memory system with hybrid search.
Includes a three-tier security sandbox for isolation.
Supports 10+ channels like Slack and Discord.
Requires Python 3.12+ and Git LFS for installation.