Search

Local-first knowledge system with sqlite-vec + ONNX embeddings. No cloud, no Docker, no PyTorch required.

Token-aware text chunker for RAG ingestion. Sentence-respecting, overlap-friendly.

Sub-millisecond PII redaction for prompts before they reach an LLM.

Low-level Python ergonomics for AWS Bedrock + Anthropic Claude: retries, cost ledger, streaming-error normalization.

Fast bulk tokenizer + token counter for OpenAI BPE encodings.

Build clean, structured LLM context files from your codebase.

AI browser automation bridge for Chinese LLMs, custom model providers, and OpenAI-compatible endpoints

Boruna: a deterministic, capability-safe programming language and framework for LLM-native applications