{
  "id": 16635124,
  "title": "The Git Forge for Agent-Human Collaboration",
  "url": "https://urgent.news/2026/10/26/the-git-forge-for-agent-human-collaboration",
  "topic": "tech",
  "section": "Tech",
  "published": "2026-10-26T17:23:11.000Z",
  "source": {
    "name": "Hacker News",
    "slug": "hacker-news",
    "url": "https://entire.com/blog/the-git-forge-for-agent-human-collaboration"
  },
  "original_language": "en",
  "account": "The \"Git Forge for Agent-Human Collaboration\" is a newly introduced platform that aims to revolutionize the way developers work with agents. The platform, named Entire, is built upon three key primitives. It fosters an open ecosystem, allowing developers to select the most suitable tools for their needs.\n\nInitially, Entire is partnering with Buildkite and Depot for Continuous Integration (CI), ensuring that build and test pipelines keep pace with the speed of agent-driven development. Entire is open to all developers, with a particular focus on AI-native teams and their agents.\n\nIn the platform, pull requests are transformed into a linear timeline where developers post comments. Continuous Integration (CI) provides a status update at the bottom of the timeline, followed by a merge button. This continuous cycle of coding and code review can clutter the timeline with lengthy descriptions.\n\nTo address this, the platform introduces \"Trail,\" which consolidates relevant information at the top of the page. It calculates evaluation scores (eval scores) after every push, providing a comprehensive overview of the development progress. The goal is to enable developers to make informed decisions about merging code based on this information.\n\nThe Trail lifecycle consists of three steps: initiation, review, and merge. A Trail can begin from Entire by filing a bug report or a feature request, or it can start directly from an agent before or after the implementation begins. Trails can span across multiple repositories, grouping together changes that belong to the same project. When code is pushed to a Trail, Entire triggers review agents to evaluate the commit.\n\nThese review agents operate independently in cloud sandboxes, providing a score and rationale for each change. For example, the platform uses Security, Slop, and Human Review Required evals. If the Slop score is orange, the agent is instructed to continue correcting and pushing code until the score turns green.\n\nEach project on Entire comes with predefined agents for code quality, security, and slop, which can be enabled for individual repositories. Developers can also configure their own agents through the Entire user interface or by using a JSON config in the repository. The platform utilizes popular AI models like Claude Code and Codex, running in isolated cloud sandboxes. Every push triggers agents to return findings and suggested fixes, which can be addressed or dismissed by agents or human developers.\n\nTo ensure that agent findings do not go unnoticed before merging, \"Gates\" can be enabled per repository. These gates define the prerequisites for approval, including human approvals and passing CI checks. Continuous Integration, a practice first proposed by Grady Booch in 1991, is crucial for modern software development, enabling frequent deployments. Entire offers first-class integrations with leading CI providers, such as Buildkite and Depot, to ensure reliable and fast software delivery.\n\nBefore the rise of distributed version control systems like Git, repositories and their history were managed on systems like Subversion. Git allowed for local commits, branches, merges, and history rewrites. However, the centralized model of Git has limitations in terms of scalability, resilience, and control over code storage. Entire aims to address these issues, providing a more robust platform for agent-human collaboration in software development.",
  "summary": null,
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}