Start with the job, then find an open-source answer

Klavis gives developers SDK and API building blocks for connecting agents to external services through MCP. Applications can create per-user server instances, combine services such as Gmail and Slack behind a Strata endpoint, or use individual integrations from a catalog of more than one hundred options with OAuth support. Open-source components can also run locally through Docker or a pipx-installed Strata package. Calls to the hosted Klavis service require an API key.

Streambert is a cross-platform desktop app for playing movies, television series, and anime from third-party sources while managing subtitles, watch history, saved titles, and downloads. Its download workflow resolves streaming playlists and relies on a separate downloader plus ffmpeg, and first launch requires a free TMDB access token. The app itself is presented as ad- and tracker-free, but users remain responsible for the legality of external content sources and downloads in their region.

GodMode is a dedicated AI browser that places full web applications such as ChatGPT, Claude, and Perplexity in resizable side-by-side panes. A global shortcut opens the window, and one input can be submitted to every service while each webview remains independently usable. It is useful for comparing web-only features, but users must sign in to the services and Windows and Linux support is still less polished.

angular-electron is a starter project pairing Angular 22 with Electron 42 for desktop development, including a development server, renderer hot reload, Vitest and Playwright tests, and executable packaging for Linux, Windows, and macOS. Hot reload applies only to the renderer, so the Electron main process must restart, and the project explicitly recommends npm rather than Yarn.

Blockbench is a free model editor focused on low-poly geometry and pixel-art textures. Artists can export models to standard formats for sharing, rendering, 3D printing, or game engines, and use dedicated formats with Minecraft Java and Bedrock-specific features. It is available as both a desktop application and a web app, with JavaScript plugins for extending the editor.

Xplorer combines multi-tab and split-pane file browsing with archive handling, rich previews, Git operations, a terminal, and SSH browsing. Users can connect their own AI provider for semantic search, file-aware chat, and categorization. The documented version is the next-branch rewrite with a new extension system; it is actively developed but explicitly not production-ready yet.

This collection combines Anthony Fu’s opinionated project and design conventions with skills generated from official Vue, Nuxt, Vite, Vitest, UnoCSS, pnpm, and related documentation. Vendored skills extend the set to adjacent tools, while git submodules help track upstream sources. It targets modern Vue/Vite work and remains a proof of concept that has not been fully tested in practice.

LobsterAI is an OpenClaw-based desktop agent that uses local tools for spreadsheet analysis, document and presentation creation, web research, browser automation, video work, and scheduled tasks. It ships with 28 configured skills, supports additional MCP servers, and can receive commands from WeChat, Feishu, DingTalk, or Telegram for remote desktop workflows.

holaOS is a local-first desktop work agent that can connect, with permission, to files, browser state, and workplace services such as Gmail, Slack, Notion, GitHub, and Jira. It summarizes those signals into durable Markdown memory stored on the machine, with SQLite vector search and RAG for retrieval. Users can inspect, edit, or delete remembered context while managing tasks, files, and outputs in one workspace. macOS is supported now; Windows and Linux are still in progress.

RuoYi AI is a full-stack platform for enterprise agent applications. It connects multiple model providers, builds local RAG knowledge bases, and lets teams compose model calls, email, human review, and other steps in a visual workflow. Its agent layer supports multiple decision modes, multi-agent coordination, MCP, and Skills extensions. A Docker Compose setup can start the backend, administration UI, user UI, and supporting services together.

Install one Python package to get both a command-line client and an MCP server for NotebookLM. You can create notebooks, add URL, text, Drive, or file sources, run queries, start web or Drive research, generate Studio content, and download resulting artifacts. Batch operations, cross-notebook queries, sharing, and Drive-source synchronization make it useful for scripts and AI assistants that need more than the web interface.

Manta is a desktop invoicing app for freelancers and people who occasionally need invoices, estimates, or receipts without adopting a full accounting service. Users can enable or hide form fields, reorder line items by dragging, customize templates and statuses, and export a PDF for printing or email. Financial data stays on the local machine, but the project describes itself as beta software rather than a complete bookkeeping system.

Altair GraphQL Client is a cross-platform GraphQL client IDE for debugging queries and GraphQL implementations. It is available through desktop applications and browser extensions, giving developers a dedicated interface for investigating query-related issues. Its role is client-side development and debugging, not running a GraphQL server.

MineContext is a desktop assistant that builds a record of a user's digital workflow from sources such as screen captures and uploaded links, then uses that context to surface proactive insights. The project describes its data processing and storage as local-first, which suits people who want an AI companion to understand ongoing work, study, or creative activity. On first launch it installs a backend environment and requires a Doubao, OpenAI, or OpenAI-compatible local or third-party model service.

ProxyBridge redirects traffic from proxy-unaware applications through HTTP or SOCKS5 proxies on Windows, macOS, and Linux, with per-process rules for proxying, direct connections, or blocking. Its GUI and CLI suit security testing, protocol analysis, and selective routing. The implementation uses platform-level interception, so supported OS versions, kernel features, and elevated permissions vary by system.

TaskingAI is a backend platform for developing and deploying LLM-based agents. It brings models from multiple providers, tools, RAG systems, assistants, and conversation history under one console and unified API. Developers can prototype configurations in the UI, then integrate them into multi-tenant products through REST endpoints or the Python SDK. The community edition can be self-hosted with Docker Compose and requires configuration for the selected model services.

SWE-bench evaluates language models on real software issues collected from GitHub by giving each model a codebase and an issue, then testing the generated patch in a reproducible Docker environment. It is useful for comparing coding agents on actual repair work, but local evaluation is resource intensive: the project recommends an x86_64 machine with at least 120 GB of free storage and 16 GB of RAM.

DeepReasoning is a self-hostable API and chat interface that routes a request through DeepSeek R1 reasoning and then Claude for the final response, including creative or coding tasks. It returns the combined workflow through one interface and supports streaming. The project is aimed at developers who want to control deployment and credentials, and it requires both DeepSeek and Anthropic API keys.