Show HN: Organize and edit local files with AI from the browser (github.com)

Conduit is an in‑browser, open‑source tool that lets AI agents operate directly on your local file system without a native app or per‑file uploads. Built as a Next.js + TypeScript front end with a Rust → WebAssembly core, Conduit uses the browser File System Access API to provide type‑safe file tools (read/write/delete/search), fast scanning/indexing via WASM, and background processing with web workers. It exposes direct tool functions compatible with LLMs (demo uses Anthropic Claude), supports regex and content‑based search, directory operations, and file editing — all running locally with zero server dependencies and files never leaving the browser. This matters because it brings AI‑assisted code and file management into a privacy‑preserving, high‑performance browser context: Rust/WASM handles large codebases efficiently while TypeScript tooling and function‑calling integration let models perform concrete file operations. Key caveats: it relies on the File System Access API (currently full desktop support only in Chromium browsers like Chrome, Edge, Opera), no mobile support, and explicit user permission is required for sandboxed directories. Conduit is available as an Apache‑2.0 monorepo (Rust, Node.js/pnpm, wasm‑pack) and provides a foundation for local, privacy‑first AI workflows that could reshape developer tooling as browser standards evolve.