Unlimited Context
A local context and memory engine that gives a small local LLM a billion-token pool on disk, pulling back only the slice it needs while it reasons instead of enlarging its context window.
Unlimited Context is a local-first context and memory engine for Ollama and other local LLMs. Rather than enlarging a model's context window, it keeps a large pool of memory on the user's own disk and pulls back the relevant slice while the model reasons, so a small local model can stay coherent across a task that would otherwise overflow its window.
Its setup command sizes the pool and checks for a local model, and works with no daemon, network or model at all using a built-in mock model for verification. Retrieved context is capped at a fraction of the model's window with room reserved for the task and output, and a slice too large to fit is skipped rather than truncated silently. It installs with pip or npm and is Apache-2.0 licensed.