Reason Machines Open-Sources Aspen, a Git Server Built for AI Agents

Why it matters

Reason Machines has released Aspen, an open-source, Apache 2.0 Git server designed for repositories that many AI agents push to concurrently. It is an independent implementation of the design OpenAI described for Pando, its internal system built to handle agent-generated push traffic at scale. The core problem Aspen addresses is that traditional Git servers serialize pushes per repository and rebuild packs for every fetch, so server load grows with the number of clients rather than the amount of new code. Aspen instead scales writes per branch: each branch is updated with a single compare-and-swap operation in PostgreSQL, while pushed packs are stored as-is in object storage and streamed to fetches without recompression. Trunk-only advertisement and background compaction keep negotiations and fetches efficient even with very large branch counts. The project is pre-alpha, lacks releases and a web interface, and its benchmarks are self-reported, but it requires no client changes and supports deployment from a laptop to cloud infrastructure.