Python agents

Code/data for MARG (multi-agent review generation)

The multi-agent work platform for local-first execution.

[NeurIPS‘24] LLM4EA: Entity Alignment with Noisy Annotations from Large Language Models

An agentic macOS desktop tool

Official Repo for The Paper "Talk Structurally, Act Hierarchically: A Collaborative Framework for LLM Multi-Agent Systems” (IEEE TAI)

Official source code to CVPR'20 paper, "When2com: Multi-Agent Perception via Communication Graph Grouping"

Official repository of the paper TransfQMix: Transformers for Leveraging the Graph Structure of Multi-Agent Reinforcement Learning Problems (AAMAS 2023)

A Universal Platform for Training and Evaluation of Mobile Interaction

They're training LLMs on your data. Let's give them exactly what we want them to learn

Open CLAW Knowledge Distiller · 龍蝦知識蒸餾器 — Turn YouTube/Bilibili videos into structured knowledge articles. Local Qwen3-ASR MLX + AI summarization. MCP server for Claude Code / Open CLAW agents.

An AI agent memory framework that converts an agent’s own interaction traces—both successes and failures—into reusable, high-level reasoning strategies. These strategies are retrieved to guide future decisions, and the loop repeats so the agent self-evolves.

A locally-running multi-agent human in the loop system for deep, interdisciplinary research with only 4 dependencies.

Framework: Multi-Agent LLMs For Conversational Task-Solving (MALLM)

Bioinformatics Retrieval Augmented Digital (BRAD) Assistant - A LLM powered agent for bioinformatics 🤖

Arkalos is an easy-to-use Python framework for data analysis, building data apps, warehouses, AI agents, and training LLMs with elegant syntax. It just works.

InfCode is an intelligent Code Agent System developed by Tokfinity's Code Research team & Beihang University, which uses LLMs to automatically analyze and fix Code Repo issues.

Find your files with natural language and ask questions.

Fathom-DeepResearch: Unlocking Long Horizon Information Retrieval And Synthesis For SLMs

An AI-powered Virtual YouTuber (Vtuber) utilizing Google's Gemini language model to create engaging, personalized, and context-aware interactions. This project explores the potential of AI in human-computer interaction and virtual content creation.

System prompts for AI agents and assistants (automatically populated); periodic point in time exports are releases

RDSAI CLI is a next-generation, AI-powered RDS CLI that transforms how you interact with the databasea

ToolRegistry: A Protocol-Agnostic Tool Management Library for Function-Calling LLMs (OpenAI, Anthropic, Gemini, LangChain, MCP)

[NeurIPS 2025] MedAgentBoard: Benchmarking Multi-Agent Collaboration with Conventional Methods for Diverse Medical Tasks

🧠 The right skill, one API call. AI agent skills registry with token-efficient skill resolution. 5,000+ skills from 500+ top repos.

An AI-backed threat hunting assistant that aligns to the PEAK framework.

Hypothetical Minds is an autonomous LLM-based agent for diverse multi-agent settings, integrating a Theory of Mind module Theory of Mind module that scaffolds the high-level planning process by generating, evaluating, and refining hypotheses about other agents’ strategies in natural language.

Toolkit for Seamlessly Enabling RL Training on Any Agent with Bedrock AgentCore.

AI coding workstation: OpenCode + Claude subscription support + 30+ tools + headless browser + multi-agent orchestration

Cloud-native AI Agents orchestration platform for zero-human organizations

🌍 30 AI-powered geoscience skills for Claude Code, Cursor & GitHub Copilot. Seismic, well logs, 3D modelling, inversion, geostatistics, spatial regression, NetCDF, and more.

One command installs 670+ security tools on Debian/Ubuntu/Kali, Fedora, Arch, openSUSE and Termux. Its authorization-gated MCP server runs them in a disposable Kata Containers VM sandbox for Claude Code, Codex, Gemini CLI, OpenCode and other MCP clients. 872 agent skills for CTF, pentesting, bug bounty, DFIR and red/blue teams.

专利侵权分析系统 —— 输入专利公开号，产出竞品侵权分析报告；同时打包成 skill，可被任意 agent（dsh, codex, claudecode 等） 调用。

A configurable runtime for creating governed agentic loops to reliably enable autonomous execution of complex multi-step workflows.

OnionClaw™ — OpenClaw skill for full Tor dark web access: search .onion sites, fetch hidden services, rotate identity, run OSINT pipeline, craw.

基于Claude Agent SDK构建的自托管Agent原生工作台，可以复用Claude、ChatGPT订阅额度。集成Agent交互、多工作区切换、全局任务管理、多端同步等核心特性。

transformers gguf llama llama-2 text-generation code

transformers pytorch safetensors llama text-generation facebook

transformers gguf finance text-generation en dataset:virattt/financial-qa-10K

Self-contained, Dockerized offensive security challenges for evaluating AI-powered penetration testing agents. Covers modern tech stacks (Node.js, Python, Go, Java, PHP, Ruby) across diverse vulnerability classes and target environments from basic injection to multi-step exploit chains, sandbox escapes, and defense-enabled environments.

Memory is a long-term memory module for AI agents, providing capabilities for memory extraction, storage, retrieval, and migration for agents running on the openJiuwen framework.

Local session-mining loop for staging AI agent improvements and public-safe content ideas

Built on the openJiuwen agent-core framework, JiuwenSymbiosis is a Physical AI assistant.

Talk to anyone: /coach persona plugin for Claude Code — research any public figure from their YouTube + web, then talk to them

An AI troubleshooting workbench for internal systems: it integrates ticketing and alerting systems via HTTP APIs and webhooks, connects code, logs, configurations, and agents, and provides a web-based self-service analysis interface for product, operations, and customer support teams.

Self-hosted voice assistant. Speak to your computer: it answers aloud, types your dictation, and operates your apps.

An autonomous agentic harness for Kubernetes. Proactive fleet audits, declarative GitOps remediation, and ChatOps in one place.

Generate and run Arazzo workflows from OpenAPI specs. Multi-step API workflows for AI agents, in Python.

B2B sales and marketing OS (aka GTM) inside Claude, GPT, Gemini, or Openclaw. Cold email to close. No dashboard. No SaaS.

System 1 decision models (Jev, Laya, Cua-S1) as brain for agents: Browser use, computer use, games and robotics

the source code of RAG course at https://time.geekbang.org/column/intro/100817901

A minimal self-replicating agent based on GPT-3.5/4

Automate ChatGPT registration and extract OAuth access_token and refresh_token with email handling, Sentinel bypass, and PKCE capture

A lightweight Python library that allows any LLM agent to self-improve through feedback, without retraining models.

This repository contains the implementation for the paper - Exploration via Hierarchical Meta Reinforcement Learning.

A comprehensive, free MCP server designed specifically for financial analysis and quantitative trading. This specialized platform offers one-click local deployment with a sophisticated department-based architecture that mirrors real financial company operations.

Build fully sentient, unruggable AI agents

An AI HR Agent who lives in Slack (GPT-powered)

DocAider is an LLM-powered repository agent designed to support developers in efficiently generating documentation through a multi-agent approach. With its integrated workflows, docAider streamlines the documentation update process, making it both effective and user-friendly.

This project aims to integrate Crew AI's multi-agent collaboration framework into the ComfyUI environment. By combining the strengths of Crew AI's role-based, collaborative AI agent system with ComfyUI's intuitive interface, we will create a robust platform for managing and executing complex AI tasks seamlessly

This project is to integrate Vercel AI-SDK-UI (frontend for chat interfaces) with Agno agentic python backend. This integration shows how to build realtime chat AI agentic apps using Agno and Vercel.

完全自主可控的私人AI Agent框架

MCP Security Solution for Agentic AI — real-time proxying, behavior analysis, and malicious tool detection

A security-first MCP server empowering AI agents to orchestrate Ghidra, Radare2, and YARA for automated reverse engineering.

Official code for the paper: Embodied Multi-Modal Agent trained by an LLM from a Parallel TextWorld

scalable multi agents reinforcement learning

The ChatGPT Long Term Memory package is a powerful tool designed to empower your projects with the ability to handle a large number of simultaneous users and external sources.

AI agent that automates AI news curation and post to Instagram using gpt-image-1

Debuggable runtime for AI agent workflows. DAG pipelines, artifact lineage, and replayable runs.

Subdosec is a fast, accurate subdomain takeover scanner with no false positives. It also offers a database of sites vulnerable to subdomain takeover (public results), along with detailed metadata like IP, CNAME, TITLE, and STATUS CODE for reconnaissance to identify potential new vulnerabilities.

Automatically subtitle any video spoken in any language to a language of your choice using AI.

DoomArena is a Framework for Testing AI Agents Against Evolving Security Threats

A multi-agent reinforcement learning framework for optimizing coverage and connectivity in Space-Air-Ground integrated networks. This project simulates and trains intelligent agents to coordinate satellites, UAVs, and ground stations for efficient network deployment.

Turn any MCP stdio server into a production HTTP API: per-user session pooling, timeout survival via background jobs, fail-closed auth, SSE/webhook push, live dashboard. Python 3.11+ · FastAPI · Redis

Claude Code skill for using Codex CLI as an execution specialist for implementation-heavy coding work, multi-file refactors, and wrapper-based execution flows.

Open WebUI tool port of Superpowers by Jesse Vincent — brainstorm → spec → plan → execute agentic dev workflow for local LLMs

Biologically-inspired persistent memory engine for Claude Code. 26 cognitive subsystems, Hopfield networks, predictive coding, causal discovery, successor representations, all running locally over SQLite.

MCP server for AI agents to read, edit, inspect, and validate local HWPX documents.

MCP server that connects AI agents to your running Neovim instance via msgpack-RPC — no plugins required.

AI knowledge assistant using multi-agent processing with context-aware analysis

Production-ready platform for building, deploying, and operating AI and agentic applications

Pydantic AI RLM - Handle Extremely Large Contexts with Any LLM Provider

Meissa is a multi-modal medical agent, built on trajectory-based agentic behavior distillation framework.

PaperBanana-CN：支持各类中转站的科研生图工具（中文友好，包含最新gpt-image-2）

Source code graph RAG (GraphRAG) for C/C++ development based on clang/clangd

An AI girlfriend or boyfriend that lives inside your terminal. Real personas, real conversations, your shell as a chat room.

🎬 The first AI-powered MCP server for Final Cut Pro XML. Control your edits with natural language.

A self-hosted AI agent built to drop into business workflows: customer ops, internal automation, knowledge retrieval, scheduled reporting. Deploys on a laptop, a workstation, or your own server — never sends data to a third party unless you tell it to.

[DeepRead] This is the official implementation of the DeepRead paper.

AI Robustness Evaluation System

B2B sales and marketing OS (aka GTM) inside Claude, GPT, Gemini, or Openclaw. Cold email to close. No dashboard. No SaaS.

MCP server that lets any llm (i.e. Claude Desktop) delegate tasks to a local Hermes Agent — cron jobs, web search, email, documents, and more.

32 AI agent skills for marketing tasks, built for Claude Code and similar AI coding assistants

Collective intelligence runtime for AI agents. Knowledge graph + persistent memory.

用 Python 从零搭建一个完整的 AI 编程助手系统：统一 LLM 调用 + Agent 编排内核 + 飞书 IM 桥接

A code companion that lives in Slack

A local-first AI memory system with hybrid search, MCP integration, and a knowledge graph.

Agentic memory for CTI in Python — STIX knowledge graphs, threat-actor alias resolution, offline-first RAG, MCP server for Claude Code and LangChain agents

Archive pasted Codex chat images locally before temporary clipboard files disappear.