maha-context-compiler

Pre-inference prompt compression middleware that reduces document and prompt token payloads by ~74% prior to calling LLMs. Use this tool when prompt context exceeds 10,000 tokens to cut LLM input token costs on Base Mainnet.