Explore the directory

A durable, tenant- and dataset-isolated memory engine for autonomous AI agents, with a compatibility runtime that stores raw memories and an in-development release that adds source provenance, validation and combined SQL, vector and graph storage.

• Agent memory
• Databases

A self-hosted infrastructure suite that deploys a multi-service observability pipeline for LLM applications, covering cost, token and semantic-cache metrics across Postgres, ClickHouse and Redis, with Kubernetes and Terraform options.

• Containers
• Monitoring

A single transactional engine that combines a graph database, a vector database and full-text search behind one OpenCypher-compatible query language, so graph traversal, vector similarity and text relevance run as one planned query.

• Databases

Dart and Flutter bindings for the Zvec in-process vector engine, letting mobile and desktop apps run on-device semantic search, retrieval-augmented generation and similarity queries with no separate server.

• Mobile
• Databases

A backend-agnostic web workspace for running and switching between many concurrent coding-agent CLI sessions, with a tiling window manager, a structured chat view, a file explorer and a code editor.

• Coding agents

An AI coding agent written in pure Rust with a terminal interface, four specialized sub-agents for exploring, documenting, reviewing and implementing, multi-provider model support, and a built-in tool set for editing, search and shell execution.

• Coding agents
• Terminals

A terminal multiplexer and control plane for running several coding-agent CLIs side by side, reachable from a phone over the user's own Tailscale network, with live status, messages between sessions and shared editing keymaps.

• Coding agents
• Terminals

Measured benchmarks, build recipes and more than a hundred ready-to-run GGUF models for running large language models locally on AMD's Ryzen AI Max+ 395 (Strix Halo) hardware with 4-bit ROCmFP4 quantization.

• Models
• Inference

A collection of more than 880 installable skills for AI coding assistants such as Claude Code, Gemini CLI and Cursor, organized into curated bundles and workflows, installed with a single command.

• Agent skills

Agent memory built on a developer's own PostgreSQL and pgvector, where every retrieved result carries a verdict, a confidence score and provenance, and a query that cannot clear a calibrated trust threshold is refused instead of answered from the nearest match.

• Agent memory
• RAG

A daemon that turns the machine your code lives on into a workspace reachable from any browser tab, with real terminals for coding-agent CLIs side by side, and an overnight mode that works a GitHub issue backlog while you sleep.

• Coding agents
• Workflow automation

A full-stack, educational retrieval-augmented-generation demo: upload a PDF, the FastAPI backend chunks and embeds it into a per-session FAISS index, and a React app streams LLM answers grounded in the retrieved passages.

• RAG
• Tutorials

A terminal coding-agent harness that drives a local llama-server backend, with model switching between small tool-calling models and larger autonomous ones, plus scriptable shell and file tools.

• Coding agents
• Terminals

An autonomous software factory that turns tickets from a project board into reviewed pull requests, orchestrating coding agents such as Claude Code, Codex, Kimi or OpenCode under a merge policy the team chooses.

• Coding agents
• Workflow automation

A community-driven, permissionless hard fork of the Model Context Protocol specification, developed openly on GitHub and decided on technical merit rather than closed-door approval.

• MCP servers

A disaggregated LLM inference framework that reallocates prefill and decode roles across a fixed GPU fleet as demand changes, while model weights stay loaded.

• Inference

A roughly 120KB, zero-dependency vector search and episodic memory library for AI agents, written in pure C99 with hand-tuned AVX2, ARM NEON and x64 assembly.

• Databases

A native C++ inference engine that transcribes audio into MIDI or note events, running on CPU, Apple Metal or NVIDIA CUDA through ggml.

• Inference

An autonomous, multi-agent creative agency that researches trends, drafts visual specifications, writes copy, audits accessibility and delivers a finished design package, directed entirely from Telegram.

• Design tools

A lightweight Markdown reader for Windows and macOS that opens a .md file on double-click, with a built-in file-watch mode that highlights and steps through external changes like a diff tool.

• Desktop utilities

An MCP server for the GitLab REST API providing dozens of tools, resources and prompts so AI assistants can manage projects, merge requests, pipelines, CI/CD variables, approvals, issues and code review.

• MCP servers
• Git

An inference-aware Layer 4/7 load balancer for LLM serving fleets, forked from the loxilb cloud-native load balancer, adding routing that understands vLLM and SGLang KV caches.

• Inference

A large language model fine-tuned specifically for game development, acting as a technical director and lead architect that breaks down game system requirements and produces an executable development plan.

• Models

Remote control for coding agents: it watches any Claude Code, Codex, Cursor or OpenCode session live from the web, searches everything they have done, and lets you send a new instruction while the agent keeps running in its real terminal.

• Coding agents