Explore the directory

An open-source tool that checks whether an AI API relay or proxy service genuinely forwards to the Claude, GPT or Gemini model it claims, by diffing probe requests against official API baselines.

• AI evals
• Security

A self-hosted AI agent operating system that keeps an agent's memory, conversations and files on hardware the user owns, running offline by default across a cluster of consumer devices.

• Agent frameworks
• Inference

A browser-based workbench that runs Claude Code, Codex, Gemini and other CLI coding agents as a coordinated local team, with an orchestrator that plans and delegates work to real terminal processes.

• Coding agents
• Agent frameworks

A zero-dependency CLI that gives coding agents a persistent, file-based project memory: a plain-Markdown protocol file that records durable decisions, requirements and constraints across sessions and machines.

• Agent memory

An AI-powered security auditing tool that uses local LLMs through Ollama or any OpenAI-compatible server to detect and analyze potential security vulnerabilities in source code.

• Security

PYAS is a source-available Windows endpoint-security application that combines YARA rules, PE machine-learning classification, digital-signature checks, optional cloud analysis and a kernel-mode driver to detect and block malware.

• Security

A graph-centric framework for orchestrating multi-agent systems that turns a natural-language intent into a visual graph design, then compiles it into an executable, traceable workflow.

• Agent frameworks

A self-hosted proxy that exposes OpenAI- and Anthropic-compatible endpoints backed by multiple LLM providers, with automatic API key rotation, failover and rate-limit handling.

• LLM gateways

A retrieval-augmented generation system that builds a hierarchical knowledge index over a document corpus, described in an EMNLP 2025 Findings paper.

• RAG
• Papers

dploot is a Python command-line tool that extracts and decrypts Windows DPAPI-protected secrets, such as saved credentials, vaults and certificates, reachable over SMB, WMI, WinRM, MSSQL or a local filesystem.

• Security

A curated catalog of MCP servers that can be discovered, deployed and run through Docker tooling, including Docker Desktop's MCP Toolkit and the Docker Hub mcp namespace.

• MCP servers
• Containers

A curated collection of papers and resources on on-policy distillation for large language models, accompanying a survey paper and a companion site with full-text search.

• Curated lists
• Papers

OSCAR is a quantization method that compresses the key/value cache used during LLM inference down to 2 bits, cutting KV-cache memory by about 8x compared with BF16 while speeding up decoding and throughput.

• Inference

A Go library for embedded vector search and semantic embeddings, with an exact cosine-similarity index and optional local (llama.cpp) or cloud (OpenAI-compatible) embedding providers.

• RAG
• Libraries

A local-first Agent OS that lets you drive Claude Code, Codex and OpenCode from a browser or chat app while your code, keys and data stay on your own machine.

• Coding agents

A self-hosted AI automation platform that builds a readable workflow, scoped AI agents and a small internal app from a single chat description, positioned as an alternative to n8n, Zapier and Make.

• Agent frameworks
• Workflow automation

A testing library that generates and runs more than 60 kinds of tests for NLP and LLM models, covering robustness, bias, fairness, toxicity and factuality, with automatic training-data augmentation.

• AI evals
• Testing

A React Native demo app for an on-device face recognition SDK, showing offline face enrollment, one-to-one verification, and liveness detection on iOS and Android without a server round trip.

• Mobile
• Auth

An inference-native agent harness for what it calls Loop Engineering: recursive agent loops built on vLLM that account for serving, routing, context windows and prefix-cache reuse while a task runs.

• Agent frameworks
• Inference

A curated collection of papers, summaries and code on deep neural network compression, covering quantization, pruning and distillation.

• Curated lists
• Papers

Research code from Meta that jointly pretrains a retrieval-augmented sequence-to-sequence language model, combining a dense passage retriever with an encoder-decoder model for few-shot question answering and other tasks.

• RAG
• Papers

An MCP server that gives AI language models access to YouTube through ten tools for video, channel, transcript and playlist search and retrieval.

• MCP servers

A from-scratch LLM stack that distills a large teacher model into a small student and serves it efficiently, with custom Triton and CUDA kernels, sharded training and paged-KV serving.

• Models
• Inference

An MCP server that lets AI assistants control AutoCAD, GstarCAD and ZWCAD through natural-language instructions, covering drawing, editing, layers, blocks and queries.

• MCP servers