Open-Source Tool Cuts LLM Token Usage by 35% Without Code Changes

A developer has released Mova Context, an open-source tool that reduces the number of tokens sent to large language models by stripping out redundant data such as repeated log lines and noisy code comments before each API call. In a benchmark test using a 53-line log file with 48 near-identical entries, token consumption dropped from 2,737 to 1,764 per run — a 35.6% reduction — without altering the underlying codebase or context meaning. The tool uses a deterministic algorithm that runs in microseconds and requires no secondary LLM call to process the cleanup. Additional features include a Cache Layout Guard for stable prompt prefixes, a circuit breaker that halts requests exceeding a set budget, scheduled job support, multi-agent orchestration, and a terminal UI built with Bubble Tea. The project is available on GitHub and supports deployment across Windows, macOS, Linux, WSL, and Docker environments.