Monty - Rust-based, sandboxed Python interpreter you can embed in agents to execute LLM-generated code fast without spinning up containers.

Zvec - In-process vector database designed to embed directly into apps for low-latency similarity search without running a separate server.

Data Formulator - AI-assisted data visualization/workbench that lets developers generate and refine charts from data using intent + editable specs.

NadirClaw - Cost-aware LLM router that automatically sends easy prompts to cheap/local models and hard prompts to stronger models.

AG-UI - Event-based protocol + SDKs for streaming agent state/messages/tool-calls into frontend apps with a standardized format.

VectorJSON - JSON parser for LLM tool-call payloads that avoids O(n²) re-parsing so UIs can react to fields as they arrive.

Spice.ai OSS - Single-binary engine that provides accelerated SQL query, hybrid search, and OpenAI-compatible inference APIs for data-grounded apps and agents.