The latest articles on DEV Community by jarzebowsky_dev (@jarzebowsky_dev). https://dev.to/jarzebowsky_dev https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F3392441%2F51ef1018-1676-4ac8-9808-8de952393e40.jpg https://dev.to/jarzebowsky_dev en jarzebowsky_dev Tue, 02 Jun 2026 17:04:40 +0000 https://dev.to/jarzebowsky_dev/i-built-an-open-source-ai-agent-for-macos-in-rust-heres-how-it-works-29n6 https://dev.to/jarzebowsky_dev/i-built-an-open-source-ai-agent-for-macos-in-rust-heres-how-it-works-29n6 <p>I spent weeks building <strong>openMOON AI</strong> — an agent that executes multi-step tasks on your Mac, not just answers questions.</p> <p><strong>"Find the invoice email and reply I'll pay Friday"</strong><br><br> → searches Mail → reads email → drafts reply → asks approval → sends</p> <p><strong>GitHub:</strong> <a href="https://github.com/niceappspl/openmoon-ai" rel="noopener noreferrer">https://github.com/niceappspl/openmoon-ai</a></p> <h2> How it works </h2> <p>Tauri 2 + Rust backend, React frontend, Node.js MCP servers.<br> </p> <div class="highlight js-code-highlight"> <pre class="highlight plaintext"><code>prompt → agent loop → McpManager → MCP server → result → back to LLM → repeat </code></pre> </div> <ul> <li> <strong>Agent loop</strong> — multi-step, streams every action to UI in real time</li> <li> <strong>MCP host</strong> — 77 tools across 5 servers (Mail, Calendar, Finder, Spotify, Safari...)</li> <li> <strong>Security model</strong> — every risky action needs your approval, everything logged to SQLite</li> <li> <strong>Local-first</strong> — Ollama works out of the box, zero cloud required</li> </ul> <h2> Looking for contributors </h2> <p>10+ <code>good first issue</code> tickets open — new MCP servers, dark mode, voice input, token counter.</p> <p>Happy to answer questions about the Rust architecture in the comments.</p> agents opensource rust showdev jarzebowsky_dev Sun, 27 Jul 2025 20:18:00 +0000 https://dev.to/jarzebowsky_dev/how-do-your-teams-manage-ai-generated-code-without-losing-control-4fdm https://dev.to/jarzebowsky_dev/how-do-your-teams-manage-ai-generated-code-without-losing-control-4fdm <p>I’m curious how teams are adapting AI coding tools like GitHub Copilot or Cody in daily workflows.<br> We found them super fast individually — but at the team level, issues popped up:</p> <ul> <li>Code quality and architecture drifting</li> <li>Security risks from AI-added dependencies</li> <li>Misaligned features vs. Jira tickets</li> <li>Difficult code reviews from AI output</li> </ul> <p>How are you managing these as a team? Do you review AI code differently? Do you have process checks in place?</p> <p>Would love to hear your experience.</p> githubcopilot ai codequality discuss