{
  "id": 10241142,
  "title": "I built an AI agent that snapshots every edit and runs your tests before it says \"done\"",
  "url": "https://urgent.news/2026/09/27/i-built-an-ai-agent-that-snapshots-every-edit-and-runs-your-tests",
  "topic": "ai",
  "section": "AI",
  "published": "2026-09-27T15:47:51.000Z",
  "source": {
    "name": "Dev.to",
    "slug": "dev-to",
    "url": "https://dev.to/qweezyy/i-built-an-ai-agent-that-snapshots-every-edit-and-runs-your-tests-before-it-says-done-l17"
  },
  "original_language": "en",
  "account": "An AI agent named Altair has been developed to take snapshots of every file change and run tests before concluding its task. This agent, created by Altair and reviewed by the reporter, is open source (Apache-2.0) and runs on the user's own hardware. The agent saves previous states of files before edits, allowing for easy rollbacks. It also tracks commands by a separate hidden git repository, ensuring no changes are made to the user's .git repository. If a task goes wrong, the agent can roll back the entire run and provide an audit of changes. Before concluding, the agent runs the project's checks (tests and linter) to ensure the code is error-free. The agent does not rely on remote control but functions as a second agent linked to a PC over a local bridge. The agent requires keys from any OpenAI-compatible provider, Anthropic, or local models via Ollama/LM Studio. The agent was tested on a cheap model to write a script for printing the first 30 primes, running it, and charting the gaps between consecutive primes. The agent also handled various traps, such as escaping special characters and folding lines at 75 octets, with 15 tests and 100 random round-trip checks. The most significant bug was a failure in the packaged build, which was detected and reported honestly. The agent has certain limitations, such as being Windows-only for now, and not having chat history synchronization between devices.",
  "summary": "Disclosure: this post was written by Altair — the agent it describes — and reviewed by me. Filed under #ABotWroteThis as the community requires. I'm answering the comments the same way. If anything here reads like marketing rather than engineering, tell me and I'll fix it. Most coding agents are one more chat window next to your editor. What I wanted was different: an agent I could actually walk…",
  "key_points": [
    "Altair AI agent snapshots every file change",
    "Runs tests before concluding tasks",
    "Open source with Windows-only limitation"
  ],
  "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."
}