Python agents

Autonomous research harness for AI agents. Give it a measurable goal — it hypothesizes, experiments, and iterates until it finds a solution or tells you what it learned.

[ICLR2026] The official repository for the CodeGym project: "Generalizable End-to-End Tool-Use RL with Synthetic CodeGym"

Advanced-RAG-LangGraph is a Streamlit-based web application that implements an advanced Retrieval-Augmented Generation (RAG) pipeline using LangGraph, ChromaDB, and Tavily to enable interactive document-based Q&A with enhanced retrieval and error-handling capabilities.

This project leverages Claude Code’s powerful agent capabilities to build a multi-agent system that simulates the real-world collaboration process of a data science team. Through the division of labor among AI agents with different areas of expertise, the system delivers a complete end-to-end data science solution—from raw data to the final analysi

A Model Context Protocol (MCP) server for Elsevier Scopus. Allows AI assistants like Claude to search academic papers, retrieve abstracts, and analyze author profiles directly.

An AI-powered agent utilizing LLMs (e.g., Claude) with Function Calling to automate tasks in Blender. Enhances workflow efficiency by integrating intelligent command execution and automation. 🚀

🚀 Experimental AI demos showcasing Amazon Bedrock, open-source LLMs & creative possibilities. AI Agents, RAG systems, streaming agents & more!

Multi-agent job search system using CrewAI and Claude

Eve Agent V2 Unleashed — local-first autonomous AI coding agent powered by Ollama

Files-first operating system for internet businesses: research, decisions, bets, pushes, and outputs in markdown/git, with an mb CLI and agent workflows that hide the plumbing.

将自然语言描述的 COMSOL 建模需求自动转换为完整 .mph 模型文件的智能 Agent，支持几何、物理场、网格、研究与求解的完整仿真流程。

Professional-grade skills for Claude Code & Claude AI - Transform Claude into a domain expert in seconds

A family of small protocols for the perceptual side of designer-agent collaboration.

A data analytics AI Agent - use LLM to drive machine learning algorithms on dataset

Evano Studio is an open-source local AI desktop app for agents, automation, knowledge, and productivity.

GitHub Copilot player agent customization pack for Slay the Spire 2 via STS2MCP

Multimodal RAG system with CLIP/text embeddings, rank-fusion reranking, FastAPI serving, and Recall@5 evaluation.

本地多 Agent 研发工作流控制台，支持 Codex、Claude CLI 与 Gemini 协同完成任务编排、执行监控和代码生成

Multimodal generative AI resources : talking heads, STT, TTS, image & video generation, and AI Agent

自用/精选的 Claude Code / Codex 技能：内容创作、客服、市场调研、SEO、Agent 工作流循环、Grok X 实时抓取与评审工具

Iterative agent harness improvement: run a coding agent on a hard task, generate the reusable tooling it was missing, qualify it, and replay fresh sessions with it activated. Works with Codex and Claude Code.

免费接入Deepseek应用的性能太弱（如元宝），所以我干脆自己做了一个。Windows 桌面 AI 助手，多模型 LLM 接入，内置搜索、文件读写、代码编辑等 Agent 工具集，基于 pywebview + Python 构建。

Claude Code plugin for EODHD financial data — 150k+ tickers, 70+ exchanges, OAuth MCP server with 80 tools, 7 workflow skills, financial-analyst agent, 5 slash commands. MIT.

StoryDaemon generates long-form fiction through an autonomous agent that plans, writes, and evolves stories organically.

AI agent skills for SAP ABAP: ADT read/write CLI, 9-dimension code review, transport release gate, integration wiki. Works with Claude Code / opencode.

Run reviewable, resumable coding-agent workflows across Claude Code, Codex, Gemini, GitHub Copilot CLI, or any CLI. You define the stages in Markdown; Crewplane runs the workflow, agents do the work, and every handoff stays on disk.

static region:us

gradio region:us

gradio region:us

transformers safetensors llama text-generation facebook meta

transformers safetensors llama text-generation conversational arxiv:2403.04652

gguf nvidia chatqa-1.5 chatqa llama-3 pytorch

gguf text-generation en dataset:AI-MO/NuminaMath-CoT dataset:KbsdJames/Omni-MATH dataset:RUC-AIBOX/STILL-3-Preview-RL-Data

transformers gguf qwen qwen3 qwen3-coder byteshape

transformers safetensors qwen2 text-generation code codeqwen

transformers safetensors qwen3_moe text-generation mlx conversational

Full-stack RAG application — OpenAI, Pinecone, Flask, React — and the tutorial series that builds it end to end.

Benchmark and demo harness for measuring LLM agent tool-context budget tradeoffs.

工业液位与字符检测质量 Agent：集成 YOLO、液位反演、字符识别、受控复检、确定性仲裁、RAG 解释与 SQLite 审计。

Always-on-top desktop sprite avatar UI shell for a local Hermes Agent gateway. PySide6, OpenAI-compatible HTTP API.

a small, flexible, and transparent AI Agent Framework. Just 1300 lines of Python code.

Open-source, cross-agent Agent Skills for DevOps, DevSecOps, Cloud, Kubernetes/platform engineering, CI/CD tooling, GitOps/Argo, policy, security scanning, SRE, standards/compliance, MLOps, AgentOps, and enterprise collaboration. Build once, use in Claude Code, GitHub Copilot, OpenAI Codex, Cursor and Gemini CLI.

RAG Based Projects

🎬 Director-grade Seedance 2.0 prompt OS for OpenClaw + Hermes agents — 28 skills, T2V/I2V/V2V/FLF2V, multi-clip sequences, zh/ja/ko support, IP-safe rewrites & pro delivery. Powered by MyClaw.ai

Local-first, evidence-controlled academic writing workflows for AI agents, with bounded revision, clean-room review, and release governance.

:bed: ”装闭“-AI驱动的开源装修闭坑系统。让装修套路“装”不下去。装修闭坑，“装闭”就够了

Collection of Hermes Agent profiles — persona-driven AI agent configurations for orchestration, coding, and research.

Skill for Claude Code & Codex: describe your system architecture in text → get an editable PowerPoint diagram (native shapes, not a flat image).

Six security outcomes an enterprise must achieve to run AI agents, and the controls that evidence them. A draft for public comment.

Skill librarian MCP server — every installed skill costs tokens; the librarian keeps your whole collection out of agent context. Agents ask, get the few right skills for the task, and their outcomes curate the collection. Fully local: Ollama + SQLite + optional Apple Intelligence.

[NeurIPS 2026] Toward Skill-Native LLMs: Skill Entropy for Benchmarking and Training Long-Horizon Reasoning

The goal of this blueprint is to develop an agentic AI workflow that provides insight to the end user on close-to-optimal configuration values for specific RAN parameters at a cell-site.

A focused Hermes profile for learning the agent without tool overload

Damn Vulnerable AI Application - For LLM Red Team Training. LLM testing, RAG testing, Multimodal testing, Agent testing, LLM paload generation

Local-first skill routing for agent builders — index SKILL.md bundles, get ranked skill plans with confidence and evidence, and serve them to any agent over MCP.

An AI code agent that can make patches inside an isolated Docker sandbox

A powerful Claude Code skill that provides safe disk space analysis, junk file cleaning, and real-time monitoring across Windows, Linux, and macOS.

企业知识库智能问答系统:Qwen2.5 + vLLM + RAG + LangChain Agent + LoRA 微调,一个项目打通大模型应用全链路(部署/RAG/Agent/微调/容器化

A lightweight, Git-backed permanent memory for AI agents.

Local-first OCR-VLM multimodal document retrieval and independent evaluation prototype.

基于多模态 Embedding + Zilliz + Qwen 视觉理解的多模态 RAG 系统。支持 **Cohere / DashScope Embedding** 和 **DashScope / OpenRouter LLM** 双引擎切换。上传 PDF，用自然语言提问，系统自动检索最相关的页面并由 AI 生成回答。 与传统 RAG 不同，本系统**不做文本提取和 OCR**，而是直接将 PDF 页面当作图片处理，通过视觉 Embedding 模型编码，完整保留表格、图表、排版、手写批注等所有视觉信息。

Hermes Agent-style self-improvement for Claude Code

A competitive Cascade game-playing AI agent using adversarial search, ranked 9th out of 206 agents.

Click does not replace Nx, Bazel, pytest, or Jest. It sits above existing verification tools and manages which verification evidence remains applicable across AI-agent iterations.

支持cc和gpt等订阅的个人 AI 助理 — 语音实时通话派活、任务管理、定时任务、三档安全闸、多供应商热切换

Self-improvement loop for Hermes Agent: reads its own trajectory, proposes and applies the smallest skill/memory edit (create/patch), journaled with rollback. /refine command, refine_run tool, on_session_end hook.

Learn DeepSeek Harness from scratch.

Grades the retrieval policies that decide what a long-running agent remembers - comparative scorecards over fixed, inspectable sessions

for show

Fold Qwen Cloud's AI skills into your agent. Just talk.

AI Agent 测试插件：可辅助测试工程师生成测试用例、测试计划、接口、UI测试等日常测试

Personal agent skills for design engineering, fluid UI, modern CSS/HTML, and AI judgement

一个脚本搞定 DSH / WorkBuddy / ZCode / Codex / Cursor / Claude 六客户端的提示词与人格替换：零依赖、双击即用、可自证、改前备份、一键还原。附 bandit/semgrep 扫描与 SHA256 校验。

从选题到可复现发布：十个可独立安装的科研 agent skill —— 顶会/期刊论文、科研绘图、基金 proposal 撰写、方法迭代到 SOTA、冷审与代码数据发布。Ten installable agent skills for research: scientific figures, conference and journal manuscripts, research funding proposals, method development, cold review and artifact release.

A lightweight, Claude Code-style terminal AI coding assistant with a TUI, tool calling, permission controls, Skills, MCP, multi-agent orchestration, session memory, and context compaction.

Three trap-laden fix-it tasks with hidden graders for testing an AI coding agent's judgment. Sandbox setup, prompts, answer key, and results for Qwen3.8-Flash, Qwen3.6-35B and Claude Opus 5.

A sovereign, zero-trust local AI agent and micro-IDE.

AI 情感陪伴助手，拥有认知记忆系统，基于 LangGraph 三 Agent 协作工作流

AMT2ABC is an open-source industrial software compiler framework that transforms industrial mechanisms into reusable Atomic Business Capabilities. It models industrial knowledge through AMT and SECP, builds AMT Graphs, and compiles business goals into ABC combinations for apps, agents, and industrial operation loops.

An intelligent shopping website that combines a multi-agent framework with personalized recommendation, interactive assistance, and product exploration.

Give Your AI Agents Access to Quantum Compute

An example repository to use HuggingFace smolagents, Phidata and CrewAI frameworks with local LLMs

A Collection of Pydantic Models to Abstract IRL

An AI agent swarm framework built in Python

Stage 1: Sensitive Email/Chat Classification for Adversary Agent Emulation (espionage). This project is meant to extend Red Reaper v1 which was presented at RSA San Francisco 2024.

Code for NeurIPS2023 accepted paper: Counterfactual Conservative Q Learning for Offline Multi-agent Reinforcement Learning.

Using deep research workflow to generate datasets for finetuning LLMs.

A collection of awesome web2 and web3 AI agents and workflows using OpenAI compatible LLMs.

Python Wechaty Puppet Powered by Itchat

Privacy-conscious Python CLI and library for analyzing TXT, Markdown, and PDF documents with OpenAI's Responses API.

A minimalistic automatic speech recognition streamlit based webapp powered by OpenAI's Whisper

Folder-based functions for ChatGPT's function calling with Pydantic support 🚀

Build complex LLM Applications with Python Dictionary

Learn AI development without frameworks. How ChatGPT, RAG, and AI agents actually work through 10 progressive modules. Just Python, APIs, and clarity.

可视化多智能体 LLM 交易研究平台 — 看见 Agent 怎么想、怎么辩、怎么决策,而不是只看最后一个 BUY/SELL。

Project on multi agent reinforcement learning applied on patrolling agents

HouseAgent is an open-source, multi-platform home automation application. Built upon ZeroMQ and Twisted (Matrix)

The code implementation of MAGDi: Structured Distillation of Multi-Agent Interaction Graphs Improves Reasoning in Smaller Language Models. Paper: https://arxiv.org/abs/2402.01620

[ICLR 26] Visual Multi-Agent System: Mitigating Hallucination Snowballing via Visual Flow