The latest articles on DEV Community by ruvian fabrin (@ruvian_fabrin_e83e2e9a9d4). https://dev.to/ruvian_fabrin_e83e2e9a9d4 https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F2514714%2Fde247d75-6216-4464-98f5-7d17af6822c6.jpg https://dev.to/ruvian_fabrin_e83e2e9a9d4 en ruvian fabrin Sun, 23 Aug 2026 00:01:32 +0000 https://dev.to/ruvian_fabrin_e83e2e9a9d4/building-a-free-local-coding-agent-with-tauri-rust-and-vue-4ahb https://dev.to/ruvian_fabrin_e83e2e9a9d4/building-a-free-local-coding-agent-with-tauri-rust-and-vue-4ahb <p>I've been building Cerne Code — a free, open-source desktop coding agent that runs 100% locally. Think "Claude Code" as a desktop app, but free and compatible with Ollama, llama.cpp, LM Studio, and any OpenAI-compatible API.</p> <p>This post covers the architecture, key technical decisions, and what I learned along the way.</p> <h2> Stack </h2> <ul> <li> <strong>Tauri v2</strong> (Rust backend) — native Windows app, ~5MB binary</li> <li> <strong>Vue 3 + PrimeVue</strong> — reactive frontend with a polished UI</li> <li> <strong>Multiple model providers</strong> — same chat completions protocol for all 4</li> </ul> <h2> What it does </h2> <p>The agent can read, edit, and create files (with a sandbox + diff review), run commands (including background dev servers), search the web (multi-engine, no API key needed), generate documents (Excel, Word, PDF, PowerPoint), control the screen via computer_use, and delegate subtasks to sub-agents.</p> <h2> Architecture highlights </h2> <p><strong>Tool execution loop:</strong> Each turn, the model decides which tool to call. The loop runs in a background async task — the UI gets progress events via Tauri's event system.</p> <p><strong>Sandboxed editing:</strong> Every write_file/edit_file goes to a project-side sandbox first. The user approves/rejects each diff before anything touches the real file.</p> <p><strong>Pipeline Dev→QA→Analyst:</strong> The agent can implement something, then a skeptical verifier confirms with real tests before declaring success. If it fails, it loops back to the dev phase.</p> <p><strong>Multi-engine web search:</strong> DuckDuckGo + Brave + Mojeek aggregated, with dedup and ranking by consensus. No API key needed by default.</p> <p><strong>Audio input &amp; TTS:</strong> Speak instead of typing. Listen instead of reading.</p> <p><strong>Skills, Agents &amp; Personas:</strong> Reusable instruction files, specialized sessions, and custom personas with their own tone.</p> <h2> Local model benchmark </h2> <p>I tested 12 local models on tool-calling ability. The winner: gemma4-e4b-qat-mtp via TurboQuant (llama.cpp fork) — 60/60 on 6 tool tasks, 0.2s per call. Full benchmark in the README.</p> <h2> Why Tauri over Electron </h2> <p>Rust compilation is slower to iterate on, but the resulting binary is ~5MB vs Electron's 150MB+. Memory usage is also significantly lower — the app idles at ~80MB.</p> <h2> Open source </h2> <p>MIT license. Download, fork, contribute.</p> <p>🔗 <a href="https://github.com/RuvianFabrin/cerne-code" rel="noopener noreferrer">https://github.com/RuvianFabrin/cerne-code</a><br> 🔗 <a href="https://ruvianfabrin.github.io/cerne-code.html" rel="noopener noreferrer">https://ruvianfabrin.github.io/cerne-code.html</a></p> agents ai rust vue