Run local language models through a chat interface or compatible API.
Project overview
TextGen provides portable desktop builds and a web interface for loading local models, including GGUF files, and using them for chat, vision, document questions, and free-form generation. It can switch among several inference backends and expose OpenAI- and Anthropic-compatible endpoints with tool calling and MCP support. Core use can remain fully offline with no telemetry, while training, additional backends, image generation, and some extensions require the full installation.
Use a familiar VS Code-style workbench inside a lighter Tauri desktop shell.
Project overview
SideX ports the VS Code workbench architecture from Electron to Tauri while retaining Monaco editing, file management, an integrated terminal, Git workflows, themes, search, and Open VSX extension installation. A Rust backend handles filesystem access, PTY sessions, Git, indexing, and process management, while the TypeScript workbench runs in the system webview. The project is still an early release: pieces such as the extension host remain in progress, and its memory-saving claims are tested most clearly on macOS, so it is not yet a drop-in replacement for a mature IDE.
Analyze US stocks locally on macOS with live charts, scenarios, and trading journals.
Project overview
Kansoku is a local-first Apple Silicon macOS app for the full cycle of watching, analyzing, and reviewing US stocks. It combines live Longbridge data, multi-timeframe charts, locally computed indicators, event calendars, scenario analysis, strategy dashboards, and frozen journal records. Users must install and sign in to the Longbridge CLI, while model choices, keys, analysis traces, and research files remain on the machine.
Manage parallel coding agents and review their work in visual editors.
Project overview
Nimbalyst is a local desktop workspace for running multiple Codex and Claude Code sessions, plus alpha-stage OpenCode sessions, and linking them to files, tasks, and a Kanban view. Developers, product managers, and designers can review red-and-green agent edits directly in Markdown, mockups, diagrams, CSV files, data models, and code. The desktop app supports macOS, Windows, and Linux, with a mobile companion for checking sessions, replying to agents, and reviewing changes away from the desk.
Run local chat, agents, and image or video generation from one Windows app.
Project overview
Locally Uncensored combines local-model chat, coding and general agent workflows, document interaction, and ComfyUI-based image or video generation in one desktop application. Its setup wizard detects multiple local inference backends and can guide installation of Ollama and ComfyUI, while conversations may use either local or configured cloud providers. The packaged application is officially tested and supported on Windows only; macOS and Linux users must build from source, and practical model choices depend on the machine and installed backends.
Inspect Kubernetes topology, events, traffic, resources, and Helm from a local UI.
Project overview
Connect a single local binary to the Kubernetes API and browse topology, resources, event timelines, service traffic, Helm releases, and GitOps state without installing an agent in the cluster or creating an account. Radar keeps cluster data on the machine, includes built-in audit checks, and exposes processed cluster context through an MCP server while relying on existing Kubernetes RBAC for access control.
Replace and customize the Windows desktop shell around your workflow.
Project overview
Seelen UI replaces major parts of the Windows 10 or 11 desktop, including the taskbar, dock, app launcher, window manager, notifications, and virtual desktops. Its CSS and JSON theming system and plugin widgets let users reshape both appearance and behavior, from minimal setups to denser power-user layouts. Microsoft Edge and the WebView runtime are required for the application to work correctly.
Upscale low-resolution or pixelated images locally with AI models.
Project overview
Upscayl uses Real-ESRGAN models to enlarge images and reconstruct detail in low-resolution or pixelated material on Windows, macOS, and Linux. Users can also try additional models or convert their own. The desktop app requires a Vulkan-compatible GPU, and many integrated GPUs may not work. It is intended for resolution enhancement rather than correcting out-of-focus or heavily blurred images.
Download videos, audio, subtitles, and playlists from hundreds of sites.
Project overview
ytDownloader wraps yt-dlp in a desktop interface for downloading video or audio from YouTube, Facebook, Instagram, TikTok, and hundreds of other supported sites. It includes playlist downloads, range selection, subtitles, and hardware-accelerated video compression. Builds are available for Windows, Linux, and macOS; the macOS app is unsigned, requires removing its quarantine flag, and expects yt-dlp to be installed through Homebrew.
Organize files, write documents, and automate browser tasks from a local desktop agent.
Project overview
Coworker runs on the desktop to sort and rename files, draft or summarize documents, connect to services such as Notion or cloud drives, and automate browser workflows including research and form entry. Users choose which folders it can access and approve or stop actions, while model access comes from their own API keys, a supported ChatGPT sign-in, or local Ollama models.
Open and edit Figma files, then generate or inspect designs with AI and scripts.
Project overview
OpenPencil is an open-source design editor available as a desktop application or web app, with native read and write support for `.fig` and `.pen` documents. Its chat agents can create and modify nodes through a large design-tool set, while the CLI, plugin API, XPath queries, and MCP server enable headless inspection, conversion, linting, token extraction, and export. The project keeps workflows programmable and local, although its roadmap still documents Figma compatibility gaps.
Browse large Git repositories and use a desktop difftool and mergetool.
Project overview
GitComet is a local-first, cross-platform Git desktop client built to remain responsive when browsing large repositories and file diffs. It can run as a standalone GUI or register as Git's difftool and mergetool for interactive or headless diff and merge workflows. The README lists staging, commits, remotes, branches, worktrees, and multi-repository browsing as planned open-source edition scope. Git 2.50 or newer must be installed locally.
Generate usable Word, PowerPoint, and Excel files from natural-language requests.
Project overview
OfficeDex is a desktop interface for OfficeCLI that turns requests for reports, presentations, or spreadsheets directly into native DOCX, PPTX, and XLSX files rather than intermediate HTML. Users can paste screenshots or upload reference images, preview generated documents inside the app, and connect OpenAI, Anthropic, self-hosted models, or the hosted runtime. Current installers target macOS and Windows, while a managed OfficeCLI subprocess performs the actual generation work.
Record a selected area of your Mac screen from the menu bar.
Project overview
Kap starts from the macOS menu bar: choose the portion of the screen to capture, begin recording, and click the icon again to stop. Option-clicking pauses an active recording. Builds are available for Apple silicon and Intel Macs, with Homebrew installation and a plugin system for extensions.
Create AI videos from script and storyboard through generation, timeline editing, and export.
Project overview
Nomi is a local-first desktop workspace for taking an AI video from an idea to an exported MP4. It connects script writing, shot planning, image and video generation, reusable assets, and timeline editing in one project. A 3D staging area lets creators position characters and cameras, lock character identity, and define camera paths so visual intent can be passed to generation models more precisely. Users bring their own model endpoint and API key, with desktop builds for macOS and Windows; generation quality, availability, and cost depend on the connected services.
Detect and remove visible Gemini watermarks from images locally.
Project overview
GeminiWatermarkTool is a local desktop and command-line application for detecting and removing visible overlay watermarks from Gemini-generated images. It supports current Gemini 3.5 profiles with automatic legacy fallback, batch processing, and optional denoising. Full GUI and smaller CLI builds are available for Windows, Linux, and macOS, with the compact edition suited to scripts and CI. It does not remove invisible SynthID marks, and video processing belongs to a separate related project.
Monitor Claude Code sessions, tool calls, and subagent activity in real time.
Project overview
This dashboard connects to Claude Code's native hooks, records session and tool events in local SQLite, and streams updates to a browser or macOS and Windows desktop app over WebSockets. It surfaces active agents, tool usage, health metrics, notifications, activity feeds, and Kanban-style status, making it easier to observe complex subagent orchestration as it happens.
Keep the macOS input source pinned by app or browser URL.
Project overview
LockIME is a macOS menu-bar app that keeps a chosen input source active globally, for each frontmost app, or for matching browser URLs. Rules can continuously reapply the source when it drifts, or switch only once on focus. Core locking needs no system permission; URL-level enhanced rules require Accessibility access, and the app targets macOS 14 or later.