{"slug": "troubleshoot-building-a-windows-ai-assistant-that-asks-before-it-acts", "title": "TroubleShoot: Building a Windows AI Assistant That Asks Before It Acts", "summary": "Team DROS, a four-member team at Hacktoberfest Hack Day Coimbatore, built TroubleShoot, a Windows troubleshooting assistant that proposes specific fixes and requires explicit user approval before executing anything. The prototype pairs a PWA with an authenticated local Windows helper, uses gemma-4-26b-a4b-it via the Gemini API for reasoning, and constrains model output to registered tools with policy checks rather than granting shell authority. The team's stated lesson is that \"an action completing successfully and a user problem being resolved are different claims,\" and full hosted guest repair remains pending.", "body_md": "A troubleshooting assistant should do more than suggest a command. It should explain what it observed, tell you what an action will change, and check whether the result actually helped.\n\nThat is the idea behind **TroubleShoot**, a Windows troubleshooting prototype we built as Team DROS at Hacktoberfest Hack Day Coimbatore with INIT Club and IDEA Club.\n\n[Explore the website and sample walkthrough](https://troubleshoot-one.vercel.app/) · [View the source on GitHub](https://github.com/Team-DROS/TroubleShoot)\n\nWindows already has automatic troubleshooters. But many support workflows still leave users moving between settings, interpreting commands, and guessing whether a fix worked.\n\nWe wanted a clearer interaction: describe the problem, inspect relevant facts, review a specific proposal, approve or reject it, and see fresh evidence afterward.\n\nWe deliberately kept the scope small. TroubleShoot is not a tool that can repair every Windows problem.\n\nThe working interface is a PWA connected to an authenticated local Windows helper. The public website is a product introduction and sample walkthrough, not a remote-control endpoint.\n\nThe flow is:\n\nThe current production provider uses **gemma-4-26b-a4b-it through the Gemini API**. A local Ollama adapter and recorded Gemma evaluations remain in the repository, but the enabled production flow is hosted-only.\n\nGemma handles reasoning. It does not receive unrestricted authority to execute shell commands. The backend constrains proposals to registered tools, while approval and policy checks determine whether an operation can run.\n\nCloud text sharing requires explicit consent for each run. API keys stay on the server side and are encrypted using Windows DPAPI outside the repository. The helper uses loopback authentication, and cancellation remains part of the workflow.\n\nThese boundaries matter because a plausible explanation is not enough to justify changing a computer.\n\nThe repository includes **real hosted read-only Windows diagnosis evidence**, alongside unit tests, API/runtime checks, and clearly labelled synthetic UI tests.\n\nOne example is the Print Spooler. Checking that it is running establishes a service fact. It does **not** prove that a printer produced a page. Our result needs to preserve that distinction instead of reporting an unqualified printing fix.\n\nFull hosted guest repair and live browser installation remain pending in the current documentation. Production desktop mutation and hosted vision are disabled. The website walkthrough uses sample data and performs no AI inference or changes to your device.\n\nThat is the most useful lesson from this build: **an action completing successfully and a user problem being resolved are different claims.**\n\nThe application is MIT-licensed. Model weights, dependencies and Windows installation media retain their own applicable terms.\n\nOur four-member split covered Windows tools and VM validation, local inference and reasoning, backend and hosted integration, and documentation. Shared contracts helped connect those components without treating model output as executable instructions.\n\nWe had worked on a separate prototype before this event. This repository was independently implemented from documented requirements following user-reported organizer guidance; earlier application code and history were not imported. The repository preserves that provenance and distinguishes new implementation from third-party tools.\n\n[Watch the TroubleShoot demo on YouTube](https://youtu.be/q4sGxlhyulc)\n\nThe website is the easiest place to explore the interaction. For real Windows diagnosis, follow the helper setup instructions and keep the local helper running. Hosted inference requires internet access and a configured API key.\n\nOur next priority is validating the full approved repair path in a disposable Windows guest, including rejection, cancellation and recovery.\n\nWhat evidence would you want an AI troubleshooting assistant to show before you trusted its proposed repair?\n\n*Disclosure: AI coding assistance was used during development, and this article was drafted with AI assistance for author review. The project documentation records implementation and validation limitations.*", "url": "https://wpnews.pro/news/troubleshoot-building-a-windows-ai-assistant-that-asks-before-it-acts", "canonical_source": "https://dev.to/umasuthan/troubleshoot-building-a-windows-ai-assistant-that-asks-before-it-acts-4l57", "published_at": "2026-10-08 10:14:41+00:00", "updated_at": "2026-10-08 10:19:08.114483+00:00", "lang": "en", "topics": ["ai-agents", "ai-tools", "ai-products", "large-language-models", "ai-safety"], "entities": ["Team DROS", "TroubleShoot", "Hacktoberfest Hack Day Coimbatore", "INIT Club", "IDEA Club", "Gemma", "Gemini API", "Ollama"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/troubleshoot-building-a-windows-ai-assistant-that-asks-before-it-acts", "markdown": "https://wpnews.pro/news/troubleshoot-building-a-windows-ai-assistant-that-asks-before-it-acts.md", "text": "https://wpnews.pro/news/troubleshoot-building-a-windows-ai-assistant-that-asks-before-it-acts.txt", "jsonld": "https://wpnews.pro/news/troubleshoot-building-a-windows-ai-assistant-that-asks-before-it-acts.jsonld"}}