> ## Content Index
> Fetch the complete content index at: https://www.stirup.si/llms.txt
> Use this file to discover other available public pages before exploring further.

# Building Git infrastructure for agent-scale development
- URL: https://www.stirup.si/building-git-infrastructure-for-agent-scale-development-9ddb7731/
- Published: 2026-10-11T19:25:17.000Z
- Updated: 2026-10-11T19:25:17.000Z
- Description: GitHub is rebuilding its Git infrastructure to support the growing demands of agentic software development, focusing on scalability, reliability, and performance for high-volume repositories.
- Author: Satish Das
- Tags: Development, Stirup Research

## Rebuilding Git Infrastructure for Agent-Scale Development

GitHub is undergoing a significant architectural transformation to support the next wave of software development, driven by the rise of agentic systems. These systems, which include autonomous agents and large-scale CI/CD pipelines, are generating unprecedented volumes of Git activity. The current infrastructure, while effective for many use cases, is being reimagined to handle the demands of repositories that receive billions of commits and requests per month.

The scale of activity on GitHub has grown dramatically. In August 2026, the busiest repository on the platform saw approximately a billion requests. Total Git activity has more than doubled in the past year, reaching 473.3 billion events per month. This growth is driven by the increasing use of agents, which perform frequent commits and checkpoints, and by the expansion of CI/CD pipelines, which execute billions of jobs annually.

Supporting these workloads requires a new approach to Git infrastructure. Traditional architectures, which rely on local disk copies of repositories for both durability and read performance, are becoming bottlenecks. The current system uses a three-phase commit protocol to ensure consistency across replicas, but this approach limits write throughput as the number of replicas increases. Adding more replicas to handle read load slows down writes, creating a trade-off that becomes unsustainable at the highest activity levels.

To address these challenges, GitHub is redesigning its infrastructure with a focus on separating durability from scale. The new architecture will use Azure Blob Storage for authoritative repository data, allowing durable storage to be scaled independently of compute resources. Read capacity will be managed by lightweight workers that cache data, reducing the need for additional durable replicas. This separation enables faster writes and more efficient handling of large read spikes from CI/CD pipelines and agent fleets.

The redesign also emphasizes minimizing coordination during writes. Only the reference update portion of a push requires agreement, while other tasks such as object storage and validation can proceed in parallel. This reduces the critical path of a push, improving overall throughput. Additionally, maintenance tasks like compaction and garbage collection will be handled by dedicated workers, rather than running on the same hosts that serve live Git requests. This change ensures that maintenance operations do not interfere with the performance of active repositories.

The new infrastructure is being built while GitHub continues to operate, without requiring any downtime or changes to existing workflows. It is designed to support the most demanding workloads, including large-scale enterprise projects, open-source collaborations, and organizations running thousands of agents against a single codebase. At the same time, it preserves the reliability and control that developers and organizations depend on, including branch protections, audit logs, and repository visibility.

By decoupling storage from compute and optimizing for both throughput and reliability, GitHub is laying the foundation for a future where agent-scale development is not only possible but efficient and scalable. This transformation ensures that the platform can continue to serve all users, from individual developers to large enterprises, while meeting the demands of the next era of software development.

**Source:** [GitHub Blog](https://github.blog/engineering/architecture-optimization/building-git-infrastructure-for-agent-scale-development/?ref=stirup.si). Analysis based on publicly available information; no hands-on testing is claimed.