O

ollama

Local-first AI agent workstation & harness: multi-agent orchestrator, review-gated task board, git worktrees, policy guardrails, MCP memory, eval gates, 6-view SSE React dashboard & 5-tab TUI across Ollama, cloud APIs & local CLIs.

Open-source DeepL and Google Translate alternative for Linux: screen translator with OCR. Select text or capture any part of the screen to get a floating translation popup. Translate offline with local Ollama LLMs, or via OpenAI-compatible APIs, CLI runtimes, LibreTranslate or MyMemory. Wayland, X11, KDE, GNOME. AppImage, Flatpak. macOS (beta). GPL-3.0-only.

Open-source AI coding agent for the terminal and browser — a guardian angel for your codebase. Runs on local Ollama models (tuned for modest hardware like mini-PC iGPUs) or any cloud LLM (Anthropic, OpenAI, Gemini, OpenRouter…). A free, self-hostable alternative to Claude Code and opencode. Website: https://hexiel.tech · by JKagiDesigns LLC (https://www.jkagidesigns.com)

Local-first voice assistant for macOS: double clap or wake word, 3-tier router (regex → local Ollama → Claude). Controls the Mac, Spotify, an Android phone and an ESP32 desk lamp.

Example code for the Guard Models deep dive on kodo-digital.fr: a guard-model classifier, an article scanner, and a benchmark of guard models on French and English prompts.

模型路由器：在 Codex Desktop 和 Claude Desktop 使用 Claude、第三方 API 与自定义模型。Local LLM proxy for Codex & Claude Desktop: OpenAI Responses / Chat Completions & Anthropic Messages, Claude Code subscription routing. macOS & Windows.

a self-hosted, locally-routed LLM stack with a default-deny security posture. your models. your machine. your terms.

The Revision Plateau: When More AI Revision Makes Work Worse — multi-task Self-Refine experiment with local Ollama

Toolkit local-first de agentes IA para OpenCode. Orquesta subagentes cloud (OpenCode Go) y locales (Ollama), con memoria persistente (engram), OCR local (uocr) y generación/edición de imágenes (ComfyUI). Incluye dos agentes primary — agente-pobre (orquestador, optimización de costo) y agente-hacker-pobre (pentesting ético autorizado + auditoría de seguridad) — con skills de SAST/pentest, y un vault Obsidian de conocimiento. TypeScript (bun) + plugins de OpenCode.

Mine v0.1 is a stable local AI Web UI foundation for home use, coding support, search, and experimentation. It combines Open WebUI, Ollama, SearXNG, Caddy hostnames, Docker Compose, and Aider into a practical self-hosted AI workflow.

CyberRAG — RAG 100% local (Ollama) sobre apuntes y libros de ciberseguridad: recuperación híbrida, respuestas con citas, asistente de pentest por fases y monitor de reindex. Sin frameworks pesados.

A self-hosted AI platform: inference, tool use, browser automation, image generation, speech synthesis, transcription, object storage, agentic code execution, and more, behind a single OpenAI-compatible endpoint. One docker-compose up.

Custom agentic harness, for both empowering and restraining small local models.

Make small local LLMs reliable at multi-step agent workflows. Constrained decoding + verify-before-commit: 96.5% end-to-end vs 2.0% unguarded on a 50-step run. Zero dependencies, runs on-prem.

Entwicklung einer Microservice-basierten KI Ticket Auto-Triage Plattform, welche eingehende Trello-Tickets automatisch analysiert und der passenden Abteilung zuweist.

Offline-first local server ecosystem using Node.js/Express, MongoDB, Electron, React, React Native, and local AI via Ollama.

Deep Self Mirror (DSM) — psycho-cognitive profile from audio, your AI-chat history or psychological session transcript. Analyze ChatGPT/Claude/Gemini exports through 8 layers, 25+ models, 400+ metrics: Big Five, Dark Triad, Schwartz Values, NLP metaprograms, attachment style, cognitive biases, Jungian shadow, C4-META 27-state topology. Age/gender-adjusted norms (Soto & John 2012, N=1.3M). Three precision modes: rule-based (instant, free), local LLM (private), API LLM (maximum). Export to HTML + JSON + DOCX + Markdown. 100% local. Open source (AGPL-3.0). Built by a psychologist (15+ yr practice). Bilingual EN/RU.

Русская версия: Deep Self Mirror (DSM) — психокогнитивный профиль из аудио, истории ИИ-чатов, транскрипта психологической консультации или интервью. 8 слоёв, 25+ моделей, 400+ метрик: Big Five, Dark Triad, ценности Шварца, метапрограммы, стиль привязанности, когнитивные искажения, тень Юнга, C4-META топология. Возрастные нормы (Soto & John 2012, N=1.3M). 3 режима: rule (мгновенно), локальная LLM (приватно), API (максимум). Экспорт в HTML + JSON + DOCX + Markdown. 100% локально. AGPL-3.0. Создано психологом (15+ лет практики). Билингва EN/RU.

Antimicrobial stewardship platform — resistance surveillance, disproportionality analysis, RAG guideline retrieval · k3s · local LLM

Local-LLM pharmacovigilance platform — FAERS signal detection, MedDRA coding, ICSR drafting · k3s · RTX 5060 Ti

Shell wrapper for OpenAI's ChatGPT, Whisper, and TTS. Features LocalAI, Ollama, Gemini, Anthropic, and more.