Explore the directory

A Model Context Protocol server that gives an AI assistant read and write access to the Unreal Engine editor, covering blueprints, materials, levels, animation and more through natural-language commands.

• MCP servers

A plugin that adds a persistent, Codex-style goal mode to OpenCode, keeping a coding agent focused on one explicit objective, with a status indicator, until it is completed with evidence, marked unmet with a blocker, or cleared.

• Agent skills

A Model Context Protocol server that lets an AI assistant discover a MongoDB database's collections, infer their schemas, and run queries, aggregations and writes through a standard interface.

• MCP servers
• Databases

An MCP server that stores the results of web searches, academic papers and code exploration in a local, project-scoped knowledge graph, so an AI agent's research accumulates instead of repeating the same lookups.

• MCP servers
• Agent memory

A large bundle of skills, subagents and hooks for AI coding tools such as Claude Code and Cursor, covering setup, testing, review and other recurring development tasks across several agent hosts.

• Agent skills

A self-hosted platform that runs real security scanners such as Nmap, Nikto and SQLMap inside a Docker sandbox and uses a locally or remotely run AI model to analyze the results, rather than only suggesting commands.

• Security

A Python library of reliability metrics for AI agents in production, designed to catch behavior such as declining decision confidence or a growing pile of unapproved actions that conventional infrastructure monitoring reports as healthy.

• AI evals
• Monitoring

RepoWiki is an open-source alternative to DeepWiki that generates a structured wiki for any local codebase from the terminal or a browser, installed with a single pip install and backed by SQLite rather than Docker or a hosted service.

• Documentation tools

An open-source, on-device voice and vision layer that gives any AI model or coding agent real-time speech input and output, screen or camera awareness, and the ability to be interrupted mid-reply.

• Speech

A local-first desktop AI chat client, built with Tauri and Next.js, that connects to cloud providers such as OpenAI or Anthropic or to local models through Ollama or LM Studio, with document parsing and a local retrieval-augmented knowledge base.

• RAG
• Inference

A reference implementation and bilingual documentation of 27 core agent-architecture patterns, each placed on a two-axis map of cognitive function against execution topology, as a companion to the book Designing AI Agents.

• Tutorials
• Books

An iOS app that answers questions by combining Apple's on-device FoundationModels with a local vector-search knowledge base and real-time web search, so responses can draw on both stored documents and current information.

• RAG

A Model Context Protocol server that reads an OpenAPI specification and exposes its endpoints as callable tools, letting an LLM client discover and call a REST API without custom integration code.

• MCP servers
• APIs

An experimental Streamlit application in which a locally run Ollama model decides for itself when to search a folder of the user's documents, rather than always retrieving text on a fixed schedule.

• RAG

A set of workflow skills for coding agents such as Trae, Claude Code and Codex that walks a mathematical-modeling competition entry from problem analysis through real computation to a finished, checked paper.

• Agent skills

A local retrieval-augmented-generation server exposed over MCP that lets Claude Code, Cursor and other agent hosts search a project's documents through hybrid semantic and keyword search, with no external service or API key.

• MCP servers
• RAG

An open-source, bring-your-own-key chat client for iOS, Android and the web that talks directly to fifteen model providers or any compatible local or remote endpoint, without routing requests through a paid intermediary.

• Inference

An endpoint-detection-and-response system built for AI agents that evaluates an agent's file access, process execution and network activity against the same policy engine as ordinary endpoint events, denying anything a policy does not explicitly allow.

• AI evals
• Security

Dropshots is a Gradle plugin and library that runs Android screenshot assertions on-device, inside your instrumentation tests, instead of pulling screenshots to a host machine for separate comparison.

• Testing

Companion code for a Chinese-language book that reimplements the core algorithms behind large language models from scratch, starting from linear regression and building up to the mechanisms behind ChatGPT.

• Books

A local SQLite engine that stores a coding agent's architectural decisions and project constraints as compact episodic memory, so switching between tools such as Claude Code, Cursor or Codex does not lose prior context.

• Agent memory

A Swift package that adds large-language-model capabilities to apps built on the Spezi framework, covering local on-device inference, OpenAI-hosted models, and inference offloaded to nearby Fog nodes.

• Inference
• Libraries

A collection of practical guides for fixing specific problems encountered while running AI coding agents and customizing terminals, each with the problem, the fix, and copy-paste configuration.

• Tutorials

A local-first AI desktop assistant that runs scheduled jobs, replies through chat apps such as Telegram, WhatsApp, Discord and Slack, and connects to real accounts like GitHub and Gmail without sending data to a hosted service.

• Agent frameworks