Find the open-source solution that fits.

Atomic Chat is a desktop application for running open-weight language models locally on macOS, Windows, and Linux. Users can load models, chat through the app, and choose among supported inference engines, while an OpenAI-compatible server on localhost lets agents, CLIs, IDE extensions, and other applications reuse the same model. Hardware acceleration and model formats vary by platform; for example, the documented Linux build currently runs GGUF models only.

OpenSERP exposes an API and CLI for live results from Google, Bing, Yandex, Baidu, DuckDuckGo, and Ecosia. A single request can search several engines, merge and deduplicate their results, and optionally return cleaned text or Markdown from target pages. This makes it useful for grounding agents and RAG workflows, as well as tracking SEO rankings across regions where a Google-only service is insufficient. You can run it locally or in Docker, or use the hosted cloud service with the same public API shape.

Particle Device OS contains the firmware libraries and system components used to build software for Particle hardware. Developers can use its build guides, application firmware API, debugging support, bootloader, cloud communication modules, threading code, and hardware abstraction layers. It is primarily a firmware development source tree, and changes to LGPLv3-licensed system components carry the sharing obligations described by the project.

Hercules converts readable Gherkin feature files into automated end-to-end browser tests, using Playwright to interact with web applications and producing JUnit XML and HTML reports. It is useful for QA or business teams that want to describe behavior without writing test code. The agent can consume tools from external MCP servers or run as an MCP server that exposes test generation, execution, and result retrieval.

KnowNote turns local documents and notes into a desktop knowledge base that users can chat with, summarize, and reason over through retrieval with source traceability. It avoids Docker and server setup while allowing private or self-managed providers such as OpenAI-compatible services, DeepSeek, and Ollama. Windows and macOS are supported. The project is still at an early stage, so its core workflow works but some areas remain rough.

Tockler automatically records computer usage and working time from active window titles and idle state. Its timeline, calendar, charts, and daily, weekly, and monthly summaries show which applications were used and how time was distributed. The free desktop application runs on Windows, macOS, and Linux, making it useful for retrospective time analysis without manually starting timers.

WebLLM Chat loads and runs language models inside a WebGPU-capable browser, keeping prompts, responses, and model processing on the user’s device instead of a cloud inference server. After the initial setup and model download, it can continue offline, with Markdown rendering, image-based conversations, and a selection of built-in models. Advanced users can also connect the interface to custom models served through MLC-LLM. Browser-native inference requires WebGPU support.

camera.ui is a self-hosted surveillance platform for viewing security cameras, running on-device AI detection, integrating with smart-home systems, and extending the setup through plugins. It can be installed as a desktop app or deployed with Docker, Proxmox, or bare-metal Linux, while mobile applications provide additional access. Footage remains on infrastructure controlled by the operator rather than requiring a cloud service. The core is free and open source, but starred capabilities including continuous recording, semantic search, and push notifications require a subscription.

Stackzy is a desktop application for inspecting Android APK files and identifying the third-party libraries they contain. It uses APK decompilation tooling behind a cross-platform graphical interface, making it useful for developers and application researchers who want a quick view of an app’s dependencies. Releases are available for Windows, Linux, and macOS, but the packaged builds are marked experimental. JAR builds require Java 15 or newer, and the macOS DMG is not notarized.

ffmpeg-commander turns common video and audio encoding choices into a browser form that produces an FFmpeg command. It helps users assemble routine operations without memorizing the tool’s large option set. The generator intentionally does not cover every FFmpeg capability and makes assumptions, so its output may need adjustment for a particular version or build configuration. An optional companion service, ffmpegd, can execute submitted jobs and return live progress to the browser.

Mirobody has a C++ core that runs inside Android and iOS apps, in a desktop wrapper, or as a self-hosted server. It brings wearable, laboratory, and clinical records behind a common interface, with FHIR endpoints, long-term memory, MCP tools, and several LLM providers. Health data can remain on the phone or the operator’s own infrastructure, but enabling an external model, synchronization, or vendor integration can send selected information outside that boundary. The architecture supplies privacy and compliance controls; the deployer remains responsible for the final HIPAA or GDPR posture.

Electronegativity is a static analysis tool for Electron developers and security auditors. It parses AST and DOM structures in source directories, JavaScript, HTML, or ASAR packages to identify security-relevant configuration mistakes and implementation anti-patterns, with CSV or SARIF output available. Findings can require manual investigation by someone familiar with Electron security, and the open-source project is no longer actively maintained.

ClawBio packages pharmacogenomics, variant lookup, ancestry analysis, UK Biobank search, and other bioinformatics workflows as skills callable from the CLI, Python, or an AI agent. Many skills emit replay commands, environment metadata, and checksums for later verification. Some workflows still depend on external databases, hosted APIs, or additional bioinformatics tools.

WebWhiz crawls a product website, turns its content into searchable context, and gives you a customizable chatbot that can be embedded for customer questions without writing integration code. It supports recurring content updates, offline-message handling, and a visual builder. Self-hosting is a multi-service deployment with an API server, JavaScript and Python workers, MongoDB, and Redis, and an OpenAI API key is required.

mcp-searxng is a standalone MCP server that connects one or more existing SearXNG instances to assistants such as Claude or Cursor. It supports paginated web, news, and article searches, time and language filters, suggestions, instance capability discovery, and conversion of readable URLs to Markdown. Multiple interchangeable replicas can fail over or be queried in parallel and merged. The project is not a SearXNG plugin, and it requires at least a configured SearXNG instance URL.

Flow Browser is an Electron-based Chromium browser with multiple profiles and spaces for separating different browsing contexts. Its interface includes a sidebar, command palette, persisted tabs, customization options, and support for installing extensions from the Chrome Web Store. Builds are distributed through GitHub Releases. Several common browser functions, including fuller download handling, pinned tabs, custom search engines, and tab previews, are still listed as upcoming, so it is best suited to users comfortable with an early-stage browser.

EvoSkill is an open-source framework for discovering and evolving reusable coding-agent skills. It runs agents on benchmark tasks, analyzes failed trajectories, proposes skill and system-prompt variants, and evaluates them on held-out data. Projects can use Claude Code, Codex, OpenCode, OpenHands, or Goose with CSV question-and-answer sets or containerized Harbor tasks. It suits teams with a meaningful evaluation set and requires the selected agent runtime, model-provider credentials, and supporting environment.

dailypaper-skills collects new papers from Hugging Face and arXiv, scores them against your research interests, and produces a daily reading list grouped by priority. Selected papers can be expanded into structured Obsidian notes covering methods, experiments, equations, figures, limitations, and follow-up questions, with indexes maintained automatically. It requires Claude Code, Obsidian, Python, and Poppler; Zotero integration is optional.