{
  "id": 12110315,
  "title": "DryRunBuddy: A Zero-Spoiler Socratic DSA Coach Powered by Local AI",
  "url": "https://urgent.news/2026/10/05/dryrunbuddy-a-zero-spoiler-socratic-dsa-coach-powered-by-local-ai",
  "topic": "ai",
  "section": "AI",
  "published": "2026-10-05T08:48:31.000Z",
  "source": {
    "name": "Dev.to",
    "slug": "dev-to",
    "url": "https://dev.to/manwitha_gopa_sunkar/dryrunbuddy-a-zero-spoiler-socratic-dsa-coach-powered-by-local-ai-20ji"
  },
  "original_language": "en",
  "account": "The Hacktoberfest Weekend Challenge invited developers to build projects that support their friends. Sunkariman Gopal created DryRunBuddy, an AI-powered Socratic coach designed to help with software engineering interview preparation. Traditional approaches involve staring at LeetCode problems, getting stuck, and relying on AI for hints, which often leads to over-explained solutions and missing the learning experience. DryRunBuddy aims to reverse this by providing a privacy-first, zero-spoiler AI coach that runs locally.\n\nThe tool operates in three stages: Socratic Coach, Code Debugger, and Optimal Mentor. The Socratic Coach analyzes the user's logic, estimates complexity, identifies edge cases where their logic fails, and asks mock follow-up questions. The Code Debugger examines the user's naive code line-by-line and offers a corrected snippet focusing on the bugs. Finally, Optimal Mentor provides the optimal solution and a step-by-step dry-run of the execution trace once the problem is solved.\n\nDryRunBuddy comes pre-loaded with 10 classic DSA patterns, each containing intentional bugs, plus a Custom Problem option. It also generates a markdown cheat-sheet and allows users to export the entire session directly to their Obsidian vault.\n\nThe app is built using a lightweight, portable stack consisting of Vanilla HTML, JavaScript, and TailwindCSS. It communicates with a local Ollama server utilizing the Gemma 2 model for AI processing. The developers achieved this without any backend infrastructure, relying solely on REST API calls to Gemma 2. They employed highly rigid system prompts to ensure Gemma's responses follow strict JSON formats, adjusting its behavior depending on the current stage—Socratic, Debug, or Solution.\n\nThe main interface presents a curated list of problems and quick revision notes. The Socratic Coaching phase breaks down the intuition without revealing the solution. When debugging code, the agent provides a patched snippet highlighting the bugs. Upon solving the problem, the Optimal Mentor offers the best solution along with a detailed dry-run execution trace.\n\nDryRunBuddy was developed with privacy, cost, and control in mind. By running Gemma locally using Ollama, users ensure that no data leaves their machine, and the tool remains free to use without any monetary costs. The developers also emphasized control by allowing users to practice a curated curriculum even without an internet connection. In the future, they plan to add voice-to-text coaching, support for additional programming languages, and further enhancements to the tool.",
  "summary": "This is a submission for the Hacktoberfest Weekend Challenge: Build for a Friend ( What I Built Preparing for technical interviews usually involves staring at a LeetCode problem, getting stuck, and asking an AI for a hint. The problem? AI models almost always over-explain and spit out the entire optimal code block, completely robbing the developer of the learning process and the \"aha!\" moment. I…",
  "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."
}