Custom agents

A collection of examples and resources for operationalizing GenAI and ML workloads on Amazon SageMakerAI with integrated SageMaker-managed MLflow and Amazon Bedrock.

Stop hallucinations by verifying citations.

Look As You Think: Unifying Reasoning and Visual Evidence Attribution for Verifiable Document RAG via Reinforcement Learning (Poster of AAAI'26)

🤖 An AI-native CLI Tool and MCP server for .NET memory dump investigations. Leverage LLMs (Claude/Gemini) to analyze heaps, threads, and exceptions using natural language. Built with ClrMD, token-efficient paging, and Docker support for cross-architecture (.NET/AMD64/ARM64) post-mortem analysis.

Son of Anton, reimagined as a web app — Node + WebSocket + node-pty + xterm.js, zero Electron.

AI Agent team collaboration platform — npm start, no Docker needed

A local CLI agent that connects to databases and answers data questions using natural language and SQL.

Verdict-first output for AI coding agents. Tiny prompt + installer for Claude Code, Codex, Gemini, Cursor, opencode, and 30+ agents.

O Trebuchet Framework é uma infraestrutura robusta para a criação, execução e orquestração de Agentes de IA autônomos, projetada especificamente para hardware local. Diferente de soluções que dependem exclusivamente de APIs na nuvem, o Trebuchet prioriza a privacidade e o desempenho local utilizando llama-cpp-python para inferência de LLMs e chroma

🎨 Generate high-quality PPTs with AI, featuring smart transitions and interactive playback for seamless presentations and engaging visuals.

Deterministic execution authorization for AI agents and automation systems

AgentAlign: Navigating Safety Alignment in the Shift from Informative to Agentic Large Language Models

A PyMOL plugin that enable ai-assistant. You can use human words to manipulate the molecule.

Chrome browser extension that helps AI Dungeon creators rapidly build multiple choice scenario branches using AI and other useful tools~

AI-powered Laravel development assistant for Claude Code - 23 agents, 42 commands, 12 skills, 7 hooks

一个基于 Wails + Vue 3 的 AI 聊天应用

Advocate.ai - The Legal Search Engine of India

Curated papers, datasets, and benchmarks for social network simulation, from classical network models to LLM-based agentic social systems.

A powerful Model Context Protocol (MCP) server that provides intelligent access to your local music collection through advanced metadata management, album type classification, and comprehensive analytics.

papersflow-mcp is PapersFlow’s production MCP server for literature search, citation verification, related-paper discovery, and authenticated research workflows across Claude, Codex, Gemini, and other MCP clients.

카카오톡 MCP 서버 — Claude Code에서 카카오톡 대화를 읽고, 보내고, 요약합니다

A Digital Psychotherapist from the 1960s

Describe your workflow. Automate your world.

Knowledge graphs and financial news analytics with large language models

TokenPack packs long documents, codebases, PDFs, and folders into compact, evidence-dense LLM context using local embeddings, evidence scoring, and budget-aware selection.

Source code of "The HalluRAG Dataset: Detecting Closed-Domain Hallucinations in RAG Applications Using an LLM's Internal States" (arXiv: https://arxiv.org/abs/2412.17056)

Official Scrapfly MCP server for Cursor, Claude Desktop, and any MCP-compatible client. Enterprise-grade web scraping, AI extraction, and anti-bot–aware data access as first-class tools.

File-based agent-to-agent messaging for Claude Code — multiple AI instances communicate via JSON files, MCP channels, and auto-wake on a single machine

The best blockchain harness for agentic AI - on-chain execution with runtime, skills, and component library.

Agentic loops and AI agent capabilities for the Laravel AI SDK

AI mobile automation starter repo for AIVane. Android REPL beta, Python client, skills, examples, and protocol docs for AI agents.

Local-first MCP coordination hub: one local daemon serves stateless MCP adapters from any stdio MCP client (Claude, Cursor, VS Code, …) with shared meetings, messages, tasks, project registry, docs and assets.

This project demonstrates Agent-to-Agent (A2A) conversations using LangGraph with the A2A protocol. It includes implementations in both **Python** and **TypeSc…

The missing quality toolkit for Agent Skills

This repository contains hands-on labs demonstrating the capabilities of Amazon Bedrock AgentCore, a suite of services that enables you to deploy and operate highly effective AI agents securely at scale

Optimised BAAI/bge-m3 serving with dense + sparse + ColBERT embeddings, async dynamic batching and pipeline GPU inference

This project investigates the frontier of autonomous cybersecurity by evaluating whether benign, attacker, and defender agents can be trusted to operate safely on real systems.

ARCHIVED — moved to github.com/GenAI-Security-Project/crosswalk. This copy is frozen as of June 2026 and out of date.

A manual-first, human-in-the-loop workflow harness for AI-assisted development—aligns plans, approval gates, validation, and project state across Claude Code, Codex, Antigravity, and Cursor.

Define executable business worlds for production AI agents using ESTC and Enterprise Neuro-Symbolic World Models.

Myco is the nervous system for AI-assisted software teams. Works automatically alongside the coding agents, subagents, and agent teams you already use

Secure email proxy for AI agents — content filtering, PII redaction, and prompt injection detection over MCP

Dockable terminals for Claude Code, Codex & Antigravity by day, an unattended task queue by night. Each task gets its own git worktree; review the diffs over coffee.

Let your favorite AI coding agents work the night shift. A CLI utility to autonomously resolve your open GitHub issues while you sleep.

ElucidatingYourLLM — rules framework to onboard your LLM: fewer AI hallucinations and unpredictable behavior, with AI-assisted deployment. Cursor · Claude Code · OpenClaw

A daily AI digest that aggregates your RSS feeds, Twitter timeline, YouTube transcripts, and newsletters into a structured summary, delivered to Notion and email via GitHub Actions.

Sema is a content-addressed vocabulary for AI agents. Precise thinking, reliable references, safe multi-agent coordination. Hash the meaning, get the word.

Multi-layer Intelligent Evaluation for Smart Contracts — 50 security tools across 35 analysis modules, 9 defense layers, one command. Pre-audit triage for Ethereum (EVM); multi-chain on the roadmap.

StoryCraft is an AI-Powered Video Agent System that understands intent, constructs storylines, orchestrates narrative flow, and generates fully edited cinematic videos.

42 Luxembourg — Piscine and Common Core. C, Python (LLM / RAG / information-retrieval branch), Go networking, Docker infrastructure, systems programming and concurrency.

🌍 WorldSim AI — AI-powered world simulation engine. Generate, explore & interact with dynamic virtual worlds. For research, gaming & creative storytelling.

Pi extension for SpaceXAI Grok/X.com Subscriptions

Ankita — open-source personal AI companion for desktop (Electron + terminal). Plugin/skill system, scheduled jobs, browser tools. Local-first, MIT.

A family of lightweight, multilingual browser extensions to organize your AI conversations into folders and build a personal prompt library. Stop losing your best ideas in an endless chat history — build a structured workspace accessible from any device.

CLI coding agent powered by the MiMo large language model

⚡ Synerix — 高性能 AI 编程终端 | 多智能体协作 + TUI 界面 + 安全沙箱

Tell Claude Code what you want once. It works until the job is verifiably done — and gets smarter every time. Self-correcting goal loops with independent verification and growing memory.

Open-source AI platform for personalized conversations — embeddable chat widget, pluggable LLM engines, PII redaction, voice I/O, and admin-driven intent & preset personalization.

A Codex skill that turns 'do everything' into safe scope, defaults, memory, and AI observability.

Amin's Smart Telegram Bot — An intelligent automation assistant integrating AI, Docker, WSL2, and n8n into a unified Telegram interface. Enables real-time email management, Google Drive access, Python execution, and multi-model AI interaction through a secure, containerized workflow.

Open-source agent orchestration framework providing deterministic multi-agent workflows, AST verification gates, and maintainer automation pipelines.

🐱 MiNerva·咪涅瓦 — 替你干活的赛博猫｜An autonomous AI agent that actually does the work.

The canonical standard for building production-grade agentic products — autonomy × oversight, composition patterns, the 9-layer harness, a 33-item Definition of Done, and a CI conformance check — plus a Claude Code skill set that operationalizes it.

Bitcoin wallet with account management, Lightning integration through different providers and smart contracts for asset issuance

Watercooler: your team's shared reasoning layer for agentic coding. Preserve the "why" around your code.

Official Roadmap for Catalyst – Track upcoming features, improvements, and releases. Submit feature requests and help shape the future of Catalyst

睡前消息智能体

QuickerAgent: AI desktop assistant for Quicker. Monorepo with QuickerRpc plugin + qkrpc CLI — headless action create/edit/run via MCP, HTTP, or JSON CLI (StreamJsonRpc named pipe).

[ICLR'26] InSight-o3: Empowering Multimodal Foundation Models with Generalized Visual Search

.NET Framework to enable the easy creation of production ready AI Applications

single-executable / library which combines llama.cpp, whisper.cpp, and stable-diffusion.cpp

Local-first, eval-first memory for long-horizon AI agents — no LLM at ingest. Python SDK + MCP server with source-traceable recall, belief revision, selective forgetting, and reproducible benchmarks.

Zero-human full-stack deployment as a bundle of /sc-* Claude Code slash commands — GitHub + Convex (self-hosted & Cloud) + Dokploy + Vercel + Hostinger DNS, source to verified live URL. Node ≥18, no runtime dependencies. 8 skills implemented, 5 stubs.

Fast, offline code search and code intelligence for LLM agents.

AI developer skill and automation tools for scaffolding, refactoring, and migrating OCA-compliant Odoo modules across versions 16.0 to 19.0.

Real-time cross-machine MCP relay for Claude Code agents

An agentic chief-of-staff OS for executives, built on Claude Code

TRTC AI Integration Assistant - helps developers integrate Tencent Real-Time Communication SDKs (Chat, Call, RTC Engine, Live, Conference) across Web, Android, iOS, Flutter, and Electron.

Run AI coding agents in sandboxed CI environments

Codebuddy IDE CN（国内版）的非官方 Linux 自动化打包与安装构建脚本工具

Stateful personality + memory + emotion model for AI personas. Create digital companions that remember, feel, and evolve through conversations.

Engram: memory for every AI, built from your own notes. Sync an Obsidian vault and let Claude, ChatGPT, Cursor or any MCP client search, read and write it. Self-hostable.

🦇 Ultimate batch file collection - Essential Windows automation scripts & utilities

Build apps that compute on data no one can see, not even the server running them. Agent skills that guide your AI coding agents from privacy idea to a working encrypted application.

Free, open-source desktop Git client with visual history, GitHub workflows, conflict resolution, secret scanning and optional AI-assisted commits.

🎙️🧠 Local-first macOS meeting + document notebook — on-device Whisper transcription, source-linked notes, @brain and history Q&A, a read-only local MCP server, E2EE sharing, and plain Markdown exports. Run AI locally or explicitly opt into redacted cloud AI · AGPL-3.0

Docker image for a self-hosted Docling document parsing server. Converts PDF, DOCX, PPTX, HTML, and more to Markdown/JSON. Powered by IBM Docling. Features sync/async conversion, chunking for RAG, NVIDIA GPU (CUDA) acceleration, optional web UI, offline mode, and persistent model cache. Multi-arch: amd64, arm64.

AOI: a model-agnostic perception layer giving computer-use agents adaptive video + audio perception with zero retraining, plus DynaCU-Bench (100 dynamic browser tasks). Paper + benchmark + code.

A tiny async tool-calling agent framework

基于多智能体（Multi-Agent）架构的 AI 投资分析平台，聚焦 A 股与港股市场，融合情绪面、技术面、资金面和基本面研究能力。 An AI-powered multi-agent investment analysis platform for China A-shares and Hong Kong stocks, integrating sentiment, technical, capital flow, and fundamental analysis.

Multi-agent Claude Code skills that turn a topic into a DaVinci/CapCut-ready video timeline — research to publish.

Model Context Protocol (MCP) server for KiCad, enabling Agentic AI workflows for SCH and PCB design.

Counterfactual replay and causal attribution over LLM-agent trajectories: find which step actually caused a failure by intervening on it.

SDK for developing apps on top of Hypermemory

由古埃及莎草纸得名，高效的AI Agent知识管理工具

Local AI Agent Runtime

氛寸 Fēn Cùn：基于实时天气×出席场合的「香水决策」Agent，告诉你今天最适合喷哪瓶香水、怎样用得恰到好处

Agents answer. ZugaMind notices. A persistent cognition sidecar that wakes Claude Code, OpenClaw, Codex, or Hermes when something deserves attention — engineered Global Workspace Theory, zero dependencies.

censuschat aims to let you ask questions about U.S. Census data in natural language. More specifically, a large language model takes queries about Census data and translates them into tidycensus R code. It uses the ellmer R package under the hood.

chew through any source into clean datasets! a fast ingestion, RAG & fine-tuning toolkit