Keep Calm and Read the Friendly Manual :-)
Description:
LingBot-Map is a feed-forward 3D foundation model for streaming 3D reconstruction. It unifies geometric grounding, dense cues, and drift correction in a single streaming framework with paged KV cache attention, enabling efficient, scalable reconstruction of long sequences.
Description:
End-to-end audio-ML pipeline that diagnoses car faults from sound; scrapes audio clips, cleans audio, embeds with CLAP, and provides calibrated triage via CLI and live web app.
Description:
MaxKB is an open-source enterprise-grade agent platform that integrates Retrieval-Augmented Generation (RAG), a workflow engine, and MCP tool-use. It supports multiple LLMs, model providers, and multi-modal inputs/outputs, enabling zero-code integration with business systems.
Description:
Obot is an open-source MCP platform that hosts MCP servers, offers an MCP registry and gateway, and enables chat clients and agents that leverage MCP servers. It supports Docker and Kubernetes deployments and emphasizes security and authentication.
Description:
PixelRAG renders web pages, PDFs and images as screenshot tiles and enables retrieval by visual similarity over the screenshots. It provides a pipeline to render, embed, index and serve a FAISS index for fast visual search, enabling retrieval-augmented generation.
Description:
Kagenti is a cloud-native middleware framework for deploying and orchestrating AI agents, with security, identity, and Kubernetes tooling. It provides lifecycle, networking, security, and observability components, enabling deployment of agents and tools via A2A and MCP protocols.
Description:
Omnigent is a meta-harness for AI agents that provides a common layer over Claude Code, Codex, Pi, and user-defined agents. It enables swapping harnesses, policy-driven governance, sandboxed execution, and real-time multi-user collaboration across sessions.
Description:
Odysseus is a self-hosted AI workspace that runs locally. It offers chat with local models, an agent that can use tools, a Cookbook for model discovery and serving, and memory-powered research and notes, all designed for privacy-first, offline-friendly operation.
Description:
Lemonade is an open-source local AI server that runs LLMs, speech, image, and other models on your own hardware. It provides OpenAI-compatible APIs, a Model Manager, and embeddable binaries for offline, private AI on CPU/GPU/NPUs.
Description:
Headroom is a library and proxy system that compresses texts, tool outputs, logs, and RAG chunks before they reach LLMs, achieving 60–95% reduction in tokens while preserving content. It includes a Python/TypeScript library, a drop-in proxy, and an MCP-based memory backend with reversible CCR compression.