Explore the directory

A desktop search and preview tool for a local ComfyUI workflow library, with semantic and AI-powered search, a node-graph preview, and live web search across YouTube, CivitAI and GitHub.

• Generative media

A fully offline, privacy-first voice assistant that combines a local mixture-of-experts language model, neural text-to-speech, real-time speech recognition and an audio codec in a C++ pipeline with no Python at runtime.

• Speech

A Rust command-line tool that filters large synthetic and pretraining text datasets by sending each row through TypeSafe AI's Jev model for calibrated, rubric-based judgments instead of a slower general-purpose language model.

• Models
• Data engineering

A stateless PyTorch optimizer that normalizes gradients by their per-slice dynamic range instead of maintaining running moment statistics like Adam or RMSprop.

• Models

A KV-cache management framework built on nano-vllm that compresses the key-value cache during LLM inference to improve memory use and throughput in high-concurrency, long-context serving.

• Inference

A Python graph execution engine for building and running agentic AI workflows, with queue-based orchestration, built-in workflow nodes and DSL import support.

• Agent frameworks

An MCP stdio server that exposes a Discourse forum's search, browsing and, optionally, posting capabilities as tools and resources for AI agents.

• MCP servers

A Rust-native, open-source alternative to Claude Code that aligns with its toolset, permission model and interaction patterns while working with Anthropic, OpenAI, DeepSeek, Bedrock or Vertex models.

• Coding agents

A set of AI coding skills that encodes a team's Android engineering standards, architecture decisions and UI conventions into Claude Code and other AI coding assistants.

• Agent skills

A Python command-line tool that finds duplicate and near-duplicate code across a codebase using tree-sitter parsing and locality-sensitive hashing, with adjustable rules for how strictly two blocks must match.

• Code quality

A 9-billion-parameter vision-language model, released with its evaluation code, that scans compressed page images and selectively expands only the relevant ones to their uncompressed form using learned tools, for answering questions over long documents.

• Models
• Papers

An Elixir and Nx implementation of a TRINITY-style small-model router that reads a conversation, extracts a hidden-state vector, and routes each turn to the right agent and role before dispatching it to a live LLM provider.

• Agent frameworks

Run-aware token budget governance for multi-agent systems, enforcing a spending limit before every model call rather than reporting cost after the fact.

• AI evals

A token-efficient MCP code intelligence server that indexes a codebase once with tree-sitter and lets AI coding agents retrieve only the symbols they need instead of reading whole files.

• MCP servers
• Agent memory

An MCP server that connects Claude, ChatGPT and other AI assistants to the Twelve Data financial API for real-time and historical prices, technical indicators, fundamentals and market news.

• MCP servers

A Godot 4.5 plugin that runs a local LLM inside a game through a bundled llama.cpp backend, giving an in-game agent node both a text reply and an action a creature or NPC can act on.

• Inference
• Libraries

A router document and a set of skills for using Claude Code, GitHub Copilot and Codex on data science projects, covering Python, SQL, data processing, visualization and statistical review conventions.

• Agent skills

An open-source enterprise agent harness that assembles company knowledge, live signals, permissions and tools into context for AI agents, and preserves their state and authority across model calls.

• Agent frameworks

A Chrome extension paired with an OpenCode plugin that lets a developer annotate any web page and send the screenshot, selected element details and an instruction straight into an OpenCode coding-agent session.

• Coding agents

A Swift library for constraining large-language-model output to a schema, regular expression or custom grammar during generation on Apple's MLX framework, built on the XGrammar constrained-decoding library.

• Inference

A daemon-backed local semantic code graph for AI coding agents, exposing MCP tools for symbol search, call graphs, impact analysis and repository context instead of repeated grep and file reads.

• MCP servers
• Agent memory

A self-hostable, open-source AI workstation that runs multiple agents and sub-agents against your own models, knowledge and tools, with a skill library, MCP integrations and sandboxed execution.

• Agent frameworks

A self-hosted, security-first AI agent with a built-in web UI, support for many model providers and chat channels, layered memory, task scheduling and MCP tool integration.

• Agent frameworks

A framework and local development server for building and scaling autonomous agents, with a project generator and an HTTP endpoint for spawning agents.

• Agent frameworks