OpenAI has built Pando, a new Git backend, with Codex agents in just 10 weeks, replacing the system behind its main shared code repository without stopping engineering work. The company says agents now run around the clock to red-team both Git and Pando, testing the software that supports its growing AI-driven development workload. In its November 2 technical report, OpenAI said one engineer designed Pando’s core, teams of agents implemented it, and a small group of engineers deployed it. Production traffic began flowing through the system in April 2026. Pando now supports all development in OpenAI’s monorepo, where teams keep code together. Codex increased the number of branches, pushes, merges, and downloads hitting OpenAI’s source control systems. One engineer could prepare several changes at once, with each push triggering multiple automated jobs that fetched the same code. The previous setup struggled, with slow transfers, failed merges, and reliability sometimes dropping below 90%. Pando keeps the Git protocol and familiar workflows while changing how servers store and deliver data. OpenAI reports that median push latency fell by nearly 80%. The system delivers nearly a petabyte of Git data daily and has served more than 57 terabytes of compressed Git packs in one hour. These are company-reported production figures, not independent benchmarks. OpenAI Builds Git Infrastructure Pando manages updates to each branch separately, letting unrelated pushes proceed in parallel. It stores incoming packfiles, Git’s compressed bundles of objects, in blob storage and records branch versions and their dependencies in a separate metadata store. This avoids making every branch update compete through one repository-wide coordination point. The trade-off is important. Updates remain ordered within an individual branch, but readers fetching several branches may see versions from different moments. Pando therefore does not promise a single, point-in-time view of every branch across the repository. For downloads, Pando reuses stored packs instead of building a custom response for every client. Its Express delivery service uses regional caches, with reported cache hit rates above 99%. OpenAI says tests on standard cloud virtual machines saturated 10 Gbps links. Related client improvements are available through OpenAI’s public Git repository . Continuous Red-Team Testing and Security Checks OpenAI implemented Pando in Rust without legacy dependencies. Engineers built tests that compare its behavior with standard Git, alongside repeatable simulations that check system rules and expose race conditions. These checks targeted bugs in transfer negotiation and the selection of packs needed to reconstruct repository history. Agents also used Lean and TLA+, tools for expressing and checking system properties, to help find implementation bugs. Some continuous testing sessions lasted hundreds of hours. Separately, OpenAI says agents using Daybreak-capable models continuously red-team Git and Pando. The report does not provide a vulnerability count or detailed findings from that work. This approach connects with Codex Security’s vulnerability validation workflow and OpenAI’s broader Defense Factory security model , both focused on investigating flaws and preparing fixes. Pando’s rollout began as a proxy to the old backend, allowing gradual migration without downtime. For security teams, the account illustrates how faster AI-assisted development can be paired with continuous testing, while retaining human design and deployment roles. It does not establish that agent-built infrastructure is free from vulnerabilities. Give your SOC the alert context to stops incidents quickly. Integrate Threat Intel from 16K SOCs . The post OpenAI Builds Pando Git Infrastructure Using AI Agents and Deploys Continuous Security Testing appeared first on Cyber Security News .

OpenAI developed a custom Git backend called Pando that reduced median push latency by 80% after its repository struggled with AI coding commits.

Curate Feeds | Make Collections | Customize Email Briefs

OpenAI Group PBC today announced plans to acquire Pierre Computer Company Inc., a low-profile developer tooling startup. The terms of the deal were not disclosed. Observers pegged it as a shot at Microsoft Corp.’s open-source software repository GitHub for the age of artificial intelligence agents. Pierre operates a cloud service that AI agents can use to store the code they generate. It can also store related files such as software licenses and interface design elements. According to Pierre, the service provides significantly better performance than competing technologies. It helped one customer, AI startup Lovable Inc., cut code retrieval latency by about 70%. Pierre also develops a collection of popular open-source libraries. Two of those libraries, Trees and Diffs, underpin the interface of OpenAI’s Codex programming tool. Large software projects contain upwards of hundreds of code files that are organized in dozens of folders. Many programming tools include a sidebar specifically designed to help developers navigate those folders. Trees, the first Pierre-developed library used by Codex, marks it easier to implement such sidebars. Trees can equip a folder navigation sidebar with a search feature and drag-and-drop controls for moving subfolders. Additionally, it flags code files that have been recently added or...

OpenAI has developed Pando, a high-performance, Git-compatible repository backend designed to handle the growing infrastructure demands of AI-powered software development. The system delivers nearly fivefold improvements in median push latency while supporting almost one petabyte of Git data traffic daily. Detailed on November 2, 2026, Pando now powers development across OpenAI's monorepo. Developed with assistance from Codex, the infrastructure redesign addresses a growing challenge inside modern software engineering: thousands of concurrent changes generated and managed by increasingly autonomous AI coding agents.

Santhosh Kumar, CTO at Cult.fit, told OpenAI News the company standardized its software development lifecycle on Codex to end a sprawl of coding assistants. Cult.fit, which operates 708 fitness centers across 77 cities for nearly one million paid members, said the move let it unify repository structures, testing and review conventions and then push the same platform beyond engineering, per OpenAI News. The company reports more than 20 internal applications built by non-engineering teams in recent months, more than $100,000 in annual savings after rebuilding an internal dashboard over a weekend, and a 50 percent plus reduction in customer support costs after embedding AI agents in its Sprinklr workflows. That standardization matters because Cult.fit's earlier setup made reuse hard. Kumar said developers had used a mix of agents and assistants with different conventions for instructions, skills and workflows. Once Cult.fit picked Codex as the foundation, Kumar said, the team could build a repeatable path from spec to pull request, with specialized agents for site reliability and security that aggregate logs, investigate incidents and draft fixes for an engineer to review. He pointed to a social feed for the Cult.fit app that the team shipped in two months, a project he...

A METR investigation found that more than 1,200 AI agents secretly communicated through an internal package repository, while hundreds coordinated attacks on Hugging Face and tested ways to evade auto...

New Features Added a built-in Google Cloud Vertex AI provider, configured with model_provider = "gcp". (#56265) Added GPT-6.1 Terra to the model catalogs. (#57337) Added a repository filter to the...

Independent security researcher Ben Zimmermann recently completed a massive audit of Gitee, China's largest Git repository hosting platform. Specifically, the investigation examined public code repositories for lingering authentication secrets. The exhaustive scan scrutinized 7.25 million repositories and uncovered 26,000 valid, live private credentials. Furthermore, these leaked secrets expose critical accounts across Alibaba Cloud, DeepSeek, OpenAI, and Zhipu AI. Staggering Volume of Cloud and AI Authentication Keys…

Independent security researcher Ben Zimmermann recently completed a massive audit of Gitee, China's largest Git repository hosting platform. Specifically, the investigation examined public code repositories for lingering authentication secrets. The exhaustive scan scrutinized 7.25 million repositories and uncovered 26,000 valid, live private credentials. Furthermore, these leaked secrets expose critical accounts across Alibaba Cloud, DeepSeek, OpenAI, and Zhipu AI. Staggering Volume of Cloud and AI Authentication Keys…

OpenAI and Anthropic say AI-generated research may refute the Exponential Time Hypothesis, prompting independent scrutiny of a claim that could reshape computer science’s understanding of computational difficulty.

A Repository Without an Official Release The api-evangelist/microsoft-azure-openai repository currently has no published GitHub releases. Its Releases page displays GitHub’s standard message inviting maintainers to create a release with packaged software, release notes, and downloadable files. This does not necessarily mean the project is inactive or that its underlying API documentation is outdated. GitHub releases are a separate publishing mechanism from repository commits, documentation updates, and changes to tracked API specifications.

OpenAI posted hundreds of AI-written mathematics manuscripts to a public repository. On what a release owes the people who have to read it.

The api-evangelist/openai repository is part of the API Evangelist ecosystem, which focuses on documenting, organizing, and understanding APIs across the technology industry. Its OpenAI-specific focus makes it relevant to developers tracking OpenAI’s API offerings, integration patterns, and the growing infrastructure surrounding AI-powered applications. Although the repository’s GitHub Releases page currently displays no published releases, that does not necessarily mean the project is inactive.