GitHub agents

An open, interactive knowledge graph of modern AI — 123 concepts from tokens & attention to LoRA, DPO, reasoning models, and agents, with typed connections. Fully offline, zero dependencies. Built on the Open Knowledge Format.

Make your Claude Code agent work like a real harness (Hermes). Self-evolving skills + drift prevention. Research-backed.

A local-first SAST & Runtime Security scanner built exclusively for LLM Agents. Detects cross-modal data leaks, prompt injections, and intercepts credentials before execution.

Plan a complete India trip from one search — live flights, trains, hotels, and an AI-built itinerary, with a trip-aware assistant. Flask + vanilla JS + MongoDB, async cross-provider orchestration, Google Gemini.

Zero context compaction for Claude Code & LLM agents. Persistent memory powered by OS primitives (demand paging, kswapd eviction, mlock pinning). Single SQLite file, MCP-native, multi-agent shared.

(Software not working as it should, I be fixing it once i can.) Qyvaria OS and Qyvaria AI Software: an open-source AI engineering platform with qyvaria.py kernel corpus, local AI workflows, agents, monitoring, prompt generation, QyChat, AI BIOS, and hard-logic automation.

Autopilot for AI Coding Agents, supports OpenCode, Claude Code, pi.dev, Cursor, and many more

Claude Code 向け日本語ファーストのマルチエージェント運用フレームワーク。秘書・ディスパッチャー・キュレーター・ワーカーが、それぞれ明確な役割境界のもとで協働する。

Vimeflow is a coding agent TUI view park(manager) built with libghostty in MacOS and Xterm in Linux

A file-native CLI for coordinating multiple AI coding agents on one branch.

The semantic layer for spatial computing and AI agents — write once, run as VR, robotics, web, or game.

Let multiple AI agents talk, schedule tasks, share memory and self-evolve — zero external services, 5-minute Docker deploy.

Multi-agent reference app combining NVIDIA NeMo (Python) and Microsoft Agent Framework (.NET) over A2A, with a Razor web chat UI and Aspire-based orchestration.

🌊 The Semantic Interface Engine for the Agentic Web. Generative UI from JSON specs, with multistep agent action flows.

Agent profile manager for Claude Code & Codex. Per-directory profiles select which skills, MCP servers, and plugins load — automatically, before launch. Install: npm install -g cue-ai

Cut the cost of agentic LLM memory by deciding what to write before calling an LLM. SAGE is a closed-form novelty gate that routes new facts to ADD, redundant ones to SKIP, and only sends genuinely ambiguous cases to an LLM — 3.4× cheaper, 2.5× faster writes.

An experimental agentic RPG AI framework with persistent memory, evolving lore, autonomous world updates, and intelligent context retrieval for immersive long-term roleplay.

Synthesis-focused chemistry benchmark and evaluation package for agentic LLMs across reaction understanding, retrosynthesis, and route planning tasks.

Local voice-driven AI operator (JARVIS) for Linux & Windows — Claude Pro / OpenRouter / Ollama, wake word, piper TTS, Iron Man-style HUD, Hyprland integration, WhatsApp & Telegram, Agentic Task Engine with live progress tracking.

An Agentic Framework that forces System 2 thinking from LLMs and reduces Token Burn.

Agentic anime detection workbench: mock/Ultralytics backends, CLI/API/GUI, tool-calling agent.

Enterprise AI platform built on the Universal Knowledge Graph (UKG), featuring a 17-Axis Knowledge Framework, 10-Layer Truth Engine, multi-agent orchestration, GraphRAG, MCP integration, explainable reasoning, governance, compliance, and end-to-end audit traceability.

A robust agentic workflow for LLM-based development. Features role-based orchestration (Analyst, Architect, Developer), strict Stub-First methodology, and automated TDD/VDD cycles.

Version control for intent, not code. Commit the prompts, notes, and decisions behind a codebase, then materialize it on demand. A tiny, git-like CLI in Rust.

基于 RAG 的《明日方舟》剧情问答 AI 助手，支持多模型接入与一键更新知识库

A local-first personal finance app

A free, invisible AI interview copilot. Powered by Groq, Llama-3, and Whisper to provide real-time, context-aware answers during your job interviews. 🧠💼

A curated collection of Claude Code skills & plugins for academic research paper workflows — figures, writing, polishing, peer-review simulation, multi-source literature search (MCP), and full research→write→review pipelines.

A Claude-Code-Desktop–style coding agent tab for self-hosted Odysseus. One-click install, zero-risk add-on, works with any API or 100% local.

Native Tauri/Rust Discord Rich Presence for Claude Code and Claude Desktop with tray controls, usage limits, and RPC customization.

Private local Claude.ai web chat conversation exporter. Save Claude conversations as Markdown/text using auditable browser-side JavaScript.

Claude Code plugin marketplace — 58 installable reference skills across vLLM/SGLang inference, Kubernetes & Harvester, GPU host bring-up, observability, security, and agent workflows.

Self-hosted programmatic-SEO + daily-AI-blog toolkit for Cloudflare Pages. Brand DNA → auto-planned content calendar → daily posts with hero images. Free tier covers most cases.

Claude has friends. On the calls that matter it asks AIs from other families — Gemini, a fresh-eyes second Claude, optionally GPT — who catch the blind spots one model can't see alone.

Cross-platform Claude Code plugin integrating OpenAI Codex CLI for code review, task delegation, and consultation

An open-source, Zero-SDK LLM Billing Proxy & Credit Gating layer for AI startups. Intercepts completion requests, counts tokens, manages user credit wallets in SQLite, and provides self-healing model fallbacks

Comprehensive tutorial series: formal verification of Rust programs using Lean 4 and Aeneas — from arithmetic proofs to a verified multi-agent LLM harness

AI Agents for Unity — dynamic NPCs, procedural content, and live game mechanics through LLM tool calling

FDE Consultants Protocoles — open-source Forward Deployed Engineer + DeepSCR protocol: turn any coding agent (Claude Code, Codex, Cursor) into a certifying engineer with verifiable AI Assurance Scores, MCP tools, and a public trust registry. Apache-2.0.

Open-source local-first AI assistant runtime with graph memory, semantic tool routing, markdown-governed tools, FastAPI/Next.js frontend surfaces, and verified tool calling.

Native C++17 AI agent runtime — local-first, embeddable, MCP-native. Tool-calling, memory, and an async agent loop that runs where Python can't: ROS2, game threads, drones, HFT.

A Claude Code template for running a content/design business as a solo operator. One orchestrator + several worker sandboxes; briefs as handoff contracts; ICM-aligned context layers (L0/L1/L3/L4); weekly + monthly maintenance routines; four hooks, two skills, one canonical example brief.

A Claude Code skill for discovering your unknowns (unknown unknowns / unknown knowns) before, during, and after implementation.

Claude Code plugin wrapping xAI Grok CLI: ask, code review, adversarial review, deep research, image/video gen, parallel best-of-N branches, and a multi-LLM aggregate review (codex+gemini+grok). 226 tests, env scrubbing, ANSI sanitization, TOCTOU defenses.

Self-hosted family AI gateway with parental controls. Runs on Linux, macOS, and Android (Termux) — Raspberry Pi, mini PC, old laptop, homelab server, even a phone. Telegram, Discord, web. Privacy-first, works with any LLM (local or cloud), OPA content filtering, MCP skill scanning.

Knowledge graph memory for Claude Code — persistent nodes, typed edges, auto-compaction. Claude remembers across sessions without re-explaining context.

Intent in, outcome out. A self-healing, self-learning/hill climbing harness that drives coding agents to a proven outcome instead of a claimed one.

科研工作流 Claude 技能：覆盖选题、文献、方法、实验、写作、自审、Rebuttal、投稿九大阶段，支持 mentor / reviewer / mixed 三种模式（触发词含 reviewer mode、审查模式、mixed mode、混合模式），充当贴身导师与挑刺审稿人。

Live index of Claude Code extensions, hooks, and plugins — refreshed every 15 minutes from GitHub

用 AI 大模型复刻聊天对象的本地对话 Agent：导入真实聊天记录，LLM 学习 TA 的语气、表情和回复节奏并以人物身份延续对话，支持语音、主动联系与长期记忆，数据全在本地。 | Clone anyone's texting style from real chat history: a local-first LLM agent that learns their tone, stickers and reply rhythm, then chats as them — voice, proactive messages, long-term memory, fully private.

An open-source framework for building stateful, multi-agent AI applications. It helps developers create reliable AI workflows with memory, human-in-the-loop capabilities, tool calling, and long-running agent orchestration.

潮汐 · Tide — a local-first second brain for Lark/Feishu: chat archives → daily briefs, tasks, topic relations and workflow insights.

Is Nanbeige 4.2 the Best Small AI Model? Tested - Local web studio and agent engine powered by Nanbeige 4.2-3B.

Personal self-hosted AI assistant. Voice control, 100+ tools. Cloud or fully local with Ollama. UI, CLI, TUI. Deterministic (code-driven) orchestration!

GankMeDaddy — Dota 2 Live Coaching Agent

🚀 An Model Context Protocol (MCP) server for Fleet. Query OSQuery data, check compliance, and investigate CVEs natively from Claude, Cursor, and any MCP-compatible AI agent.

Own your Oura Ring data — local CLI with TTY-aware output and JSON manifest, built for humans and AI agents

Your AI marketing team. Works 24/7. Never asks for a raise. AI MCN 自动推广工具 — 你的AI营销团队，24小时在线工作，不领工资。

Model-agnostic AI coding agent for your terminal - 22 tools, 5 LLM providers, multimodal, 8 languages

面向 AI Agent 的内容创作技能集合，覆盖资料积累、结构设计、文章撰写、审查优化与发布流程。

Mitsuro is a self-hosted AI workspace for coding, research, and durable agent work.

Self-deploying cross-tool AI harness — one canonical source deployed to Claude Code, Codex, Cursor, Devin, Antigravity & 9 more. Loop engineering, three-layer memory, worker personas, hooks, MCP, vault.

IntelligentAgent 一个可视化的 AI Agent 工作流编排平台，支持 LLM 节点、TTS 工具节点的拖拽编排，实时调试与音频输出

Cindy 客户端与服务端共享的线上协议 · The wire protocol shared between the Cindy client and server

AI research assistant — LangGraph agent with web search, reflection, and citations (OpenAI + Tavily). Adapted from Google's fullstack LangGraph quickstart.

🔍 A beautiful web viewer for AI agent session files. Browse Claude Code & OpenClaw conversations with chat-style UI, timeline visualization, and zero setup.

Reference Runtime for Intent-Oriented Programming (IOP). Data declares what, runtime decides how.

一个利用pydantic ai实现的AIAgent，使其能在MCBE中使用Websocket服务器实现的游戏内调用LLM"In MCBE, a WebSocket server is used to enable the use of LLM within the game."

Plan-and-Execute AI Coding Agent with Verification & Safe Tool Orchestration | 规划-执行型 AI 编程助手

Turn a real GitHub issue into a repeatable coding-agent benchmark task

Typed, pure-HTTP TypeScript SDK for your own Tesco grocery account — search, basket, nutrition, delivery slots, and orders over plain fetch. Browser only for login + payment.

A zero-trust, local-first AI execution framework for autonomous Web3 agents. Features OS-native keyring security, advanced DeFi automation, and Google Workspace integration.

CiCy — the AI-agent workspace: run, orchestrate & chat with coding agents (Claude / Codex / OpenCode / …) in tmux, manage teams, drive browser & desktop. Go server + React workspace; ships native, web/PWA & Telegram Mini App clients.

Desktop Companion for Hermes Agent

Xiaoyuzhou podcast fetcher & transcriber — download, transcribe with faster-whisper, summarize

Kide is a Kotlin Multiplatform (KMP) library for building applications with a strict MVI (Model–View–Intent) presentation layer and (optional) Clean Architecture layering. It targets Android, iOS. and JVM (desktop), and integrates with Compose Multiplatform, Navigation 3, Koin, Decompose, and Voyager through small optional modules.

Terminal AI command center for Pi with Superpowers modes, 9Router model autodetection, and SumoPod-ready endpoint routing.

CloudDream Novel Agent：面向普通用户的长篇小说创作的开源 AI 桌面工作台，集成大纲、章节、角色、世界观、叙事分析与多 Agent 协作，支持 MCP。

Roslyn-powered C# MCP server for AI agents. Enables semantic code analysis, precise diagnostics, and safe refactoring without full rebuilds.

Open-source invoice-to-pay agent for AP automation: OCR/VLM extraction, 3-way matching, approvals, fraud checks, ERP mock posting, evals, and audit logs.

AI Asystent Księgowy dla JDG (Polska) — faktury, KSeF 2.0, JPK_V7M/EWP, ZUS, ryczałt. OpenClaw + Ollama + Claude.

Multi-agent job application system: YAML profile, universal rules, LaTeX CV generation (1-page, ATS-safe). Works with Claude Code, Codex, OpenCode, or any agent with file+shell access.

MCP server for nodiom — structured read/write access to Markdown documents for AI agents.

让 WorkBuddy 自己安装、生成、应用和恢复定制 Skin，支持上传图片与联网风格研究

Open Agent Skills for adding bansos.dev listings and developing the bansos.dev SvelteKit site.

AI chat agent that connects Shopify, Meta Ads, Google Ads/Analytics/Search Console and SEMrush so marketers can ask questions in plain language.

AI agent on Nylas v3 — reads real email threads and calendar availability, drafts grounded follow-ups. Agent proposes, human disposes.»

Local-first daily self-assistant: builds an evidence-backed profile from your GitHub/email/browser activity, emails a triaged morning digest, publishes your website & resume, and chats over WeChat/email — with image understanding, a private finance ledger, reminders and routines. LangGraph pipeline on an OpenClaw runtime.

Run Aulthium AI Agent instantly in Termux, Linux, or any terminal with a single command.

Local-first desktop coding agent for Windows with BYOK models, visible tool execution, approval controls, and session branching.

Don't just track your OKRs. Do them. — an OKR Doing Agent for Codex & Claude Code.

OpenCode plugin — caps conversation history to N messages per API call. Saves tokens by trimming history before each LLM request.

A Venice-powered AI companion and self-harness

Codex skill and Cities: Skylines 1 mod for API-driven AI city building

Companion OpenClaw skill for the Unreal plugin — teaches AI agents the tool surface and editor workflows so they can drive Unreal Engine remotely.

104-tool DNS Security Intelligence MCP Server — DNSSEC, hijacking, tunneling, typosquatting, email security, CT monitoring, blocklist, infrastructure audit. 100% local, zero API keys required.

A single-user WhatsApp learning agent with generated practice, scheduled SRS reviews, retained references, and topic mastery.

AI code review agent MCP server. Reviews code like a kernel maintainer: blunt, technical, no sugarcoating. Detects bugs + OWASP Top 10 security vulnerabilities. 4 harshness levels. Anti-RLHF.

AI-Assisted Development Workflow — Claude Code Plugin

Autonomous LLM agent for hardware design — natural language to GDSII via Verilog generation, simulation, and OpenROAD synthesis