Python agents

The corresponding data and code from our paper "A Benchmark for Deep Information Synthesis" (ICLR 2026). Do not hesitate to open an issue if you run into any trouble!

A portable workflow scaffold for agentic projects, with SQLite-backed epics, tickets, scoped roles, and explicit review gates.

🔒 Private AI Engine for Data Intelligence. Automate complex workflows with agentic automation, knowledge graphs, and full data control. On-premises, GDPR-compliant.

Local AI (LLM) Ollama Tool and Extension. It's a hammer /approve /reject are your friends.

Sapti AI implements the Hive Mind Protocol — a seven-agent architecture evolving AI companion that remembers users personally, distills collective wisdom from all conversations (the Hive Mind), and measurably grows its personality over time.

Assistente virtual inteligente para WhatsApp com IA, Celery, Docker e integração com APIs modernas. Escalável, modular e pronto para produção.

MCP wrapper for Burp Suite that builds complete, well-formed HTTP requests from structured input + ingests dedup/JS exports for token-thrifty AI-assisted pentesting. Bundles a fixed Burp extension.

Claude × Penn AI Hackathon 2026 — First Place Project (Democratic Governance)

AI humanizer: CLI tool and Claude Code skill for rewriting AI-generated text into natural human writing.

Stop hitting your Claude Pro/Max 5h cap. Spread your day across more windows. Skill + Python script + claude-code-scheduler dependency.

The first MCP server for autonomous Flutter testing on real iPhones and Android devices. 110 tools across Android (uiautomator2+adb), iOS (WebDriverAgent+pymobiledevice3), Flutter (Patrol + flutter run --machine). Works with Claude Desktop, Claude Code, Cursor.

基于 ReAct 的智能助手，支持 RAG、MCP 与 Skill 扩展。

MCP server for the Aesthetics Wiki (aesthetics.fandom.com) — search, read, and discover aesthetics from your LLM.

Enable AI agents to search, crawl, and extract web data with IP rotation, CAPTCHA handling, and rate limit management via CLI and Python.

从零实现四种 LLM Agent 范式：ReAct / Plan-and-Solve / Reflection + 综合案例：电商客服智能体

OIH: autonomous LLM-agent platform for computational binder design and ADC prioritization

Synthos — 自主进化学术科研平台 / A self-evolving cognitive operating system for scientific research. 7 cognitive atoms + evolution engine. Pure SKILL.md-driven, zero Python scripts.

Multi-agent research assistant built with LangGraph. Demonstrates agentic workflows, tool use, and human-in-the-loop patterns with OpenAI and Anthropic LLMs.

TeamBench: Evaluating Agent Coordination under Enforced Role Separation

Claude Code Skill that transforms Claude into a multi-agent hedge fund analyst with 8 AI agents for comprehensive stock analysis

Per-agent scoped MCP tool proxy — credential isolation, resource scoping, and audit logging for AI agent deployments

Autonomous YouTube AI agent: LLM script, TTS voice, stock clips, FFmpeg editing, AI thumbnail, auto-upload — 100% free, Telegram-controlled.

Oracle AI Agent Harness — fork of NousResearch/hermes-agent powered by OCI GenAI and Oracle 26ai Free

The localhost AI Agent Runtime - Chat UI, Tools, RAG, and MCP in one pip install

Chinese legal assistant Agent Skill for Cursor and Claude Code. Covers legal consultation, litigation analysis, contract review, and legal research.

CyClaw is a secure, offline-first local AI agent built on three invariants: RAG-first retrieval, LangGraph topology as security policy - https://cgfixit.com/CyClaw - https://bit.ly/CyClawDemo

AI CTFer: A fully automated AI agent for solving CTF challenges.

Trustworthy memory system for LLM companions. Prevents memory fabrication with provenance tracking, write gates, and grounding audits.

LLM-powered Quantitative research assistant which combines financial data, quantitative models, and natural laungage generation to produce insightful market research reports. AI/ML + FinTech Engineering

CXWorkflow is a multi-session Codex development workflow. It does not treat “more agents” as better; it uses minimum necessary concurrency to organize Codex into a predictable, recoverable, long-running development team.

An MCP server that lets any LLM research and trade Polymarket. Point Claude, Cursor, or VS Code at it and your agent can read markets, prices, and your account — and place real trades. Optional REST gateway for algorithmic bots.

Documentation for ORMCP - An MCP Server for bridging AI and Database

An IDE for managing software projects with the Claude Agent SDK — a web cockpit + a kanban board that auto-runs cards as full-auto agent sessions.

Relationship Persona Simulation Engine — 5-layer personality model, nonlinear state dynamics, LLM dialogue analysis. Build a digital twin from chat history. 不是替代真实的人，而是帮你更理解人心。

A culinary agent for Claude Code: design and cook a dish through 7 phases, grounded in regional cooking knowledge with dynamic ingredient curation.

Multi-model panel and judge (OpenRouter-Fusion style) for Claude Code, opencode, and any MCP host. Diverse models answer in parallel, then a judge reconciles them into one better answer.

Python SDK for Dakera AI agent memory — self-hosted, 88.2% LoCoMo. Vectors, hybrid search, knowledge graphs, sessions.

Local-first AI writing control center for long-form fiction: structured story memory, long-context prompts, consistency checks, AI-tone cleanup, and IP adaptation.

A Full-Stack Open-Source Platform for Agentic AI Research Assistance via Deeply-Engineered Multi-Paradigm RAG

破窗造视 Fractovision — MiniMax模型多媒体生成引擎。图片+视频+语音+音乐四合一API封装。

Strands Agents-powered, Amazon Nova LLM-driven Mars Curiosity Rover Agent. Natural-language control of a NASA Curiosity rover & AWS DeepRacer — multi-runtime (ROS2 / rosbridge / HTTP), autonomous Nav2 + live SLAM.

Enterprise-grade schema drift detection for Microsoft Fabric with automated change detection, impact analysis, schema history, and DevOps integration.

【开发中】基于 LangGraph 图编排与事件溯源的 LLM 驱动 COC7版规则跑团守密人系统

Scan your LLM-powered applications for authorization gaps, leaked credentials, missing rate limits, prompt injection risks, and other security issues — before they reach production.

AI-powered English learning platform with FastAPI, Next.js, LangChain, OpenAI, and Azure integrations for personalised assessments, adaptive exercises, speech analysis, and intelligent feedback generation.

Turn any topic into a complete research report using AI agents. ResearchForge autonomously plans tasks, searches the web, analyzes information, and generates actionable insights.

智驭万象 · Cosmosys — 全球首款端到端多智能体控制自动化平台。Natural Language → Controller Code.

AI-powered autonomous business platform with Ray, FastAPI, and Streamlit — marketing automation, content generation, and e-commerce

A dynamic-workflow engine for any agent, any model: concurrent leaves, cost-aware routing, schema enforcement, crash-resume, bounded iterate-to-goal loops with independent verification, and shared-context dedup that cuts fan-out input tokens 70-89%.

Watches live drilling data, flags trouble before it becomes an incident, and answers plain-English questions about the well — built entirely on Equinor's real Volve field data.

Self-hosted multi-agent AI platform. Persistent labs, 40 sandboxed tools (code, web, RAG, GPU dispatcher, web3), private Qdrant + LightRAG, sandboxed code execution per lab, anti-loop detection, Ollama/vLLM/OpenAI/Anthropic adapters. Two-command Docker deploy. Apache 2.0.

🔥 LangGraph 多智能体研报系统 | Planner-Researcher-Reviewer 评审闭环 + 自动打回迭代 | FastAPI 异步任务 + Redis | Streamlit 实时看板 | DeepSeek 驱动

This repository contains code and walk-through for creating a local context engine

Run AI agents on your machine. Each agent has persistent memory, connects to your real tools, and talks to you from terminal, Telegram, Discord or Slack simultaneously. Nothing acts without your approval. Nothing leaves your machine.

AI-powered Reinsurance Contracts & Parties management

A maintainability ratchet for AI-assisted Python.

Self-hosted AI Analytics Engineer. Bridges the gap between your codebases and databases.

Four AI engineering skills that gate your agent before it builds. Evaluated with deterministic gates, behavioral probes, and LLM-as-judge rubrics.

AVE - Agentic Vulnerability Enumeration. The behavioral classification standard for agentic AI components.

A lightweight spec-driven loop harness for Claude Code, Codex, and Gemini CLI. Fresh-context runs, specs, commits, and gates to get you and your agents productive immediately.

AI Growth Mirror 是一面本地优先的 Agentic 操作成熟度评估系统。它读取 Cursor、Claude Code、Codex 等 AI 编码工具的本机会话历史，把“协作是否稳定、问题出在哪、如何改进”转化为可观察、可解释、可追踪的成长报告与训练建议。不是 Prompt 打分器，而是帮助开发者建立可复用 AI 协作系统的成长镜。

Lightweight token tracking, cost management, and budget enforcement for LLM API calls

Persistent code-knowledge memory for AI chats. An MCP server that caches a codebase's structure and summaries in a local SQLite "brain" so sessions query it instead of re-reading files — token-cheap, 100% local.

Audit your Salesforce RevOps. Then fix it. A Claude Agent Skills pack for Salesforce admins and RevOps practitioners.

A simple, elegant Python library and Mini SDK for AI models with powerful features. Built for developers who want to integrate AI into their projects without dealing with complex API setup. Supports Google Gemini, OpenAI, Anthropic, and any custom provider via the Adapter Pattern.

修复ClaudeCode更新到最新版本后，无法使用第三方模型的问题。Fix the issue where Claude Code can't use third-party models after updating to the latest version.

Critical thinking plugin for Claude Code. Three specialized agents (advocate, adversary, judge) argue, stress-test, and judge any idea using a shared argument graph, 10 reasoning traditions, and persistent memory.

Self-hosted multimodal RAG cloud memory service — a shared brain for your AI agents. Multi-agent cross-query, LanceDB + LightRAG + RAG-Anything.

Measure citation and evidence stability in RAG pipelines across runs and configuration changes.

A comprehensive, structured AI tech stack directory covering the complete ecosystem from foundation models to end-user applications. Helping developers, product managers, and decision-makers quickly navigate AI technology choices and the full landscape. | 一份全面、结构化的 AI 技术栈目录，涵盖从基础大模型到终端应用的完整生态。帮助开发者、产品经理和决策者快速了解 AI 领域的技术选型与全景格局。

Production-grade Deep Research AI with web search, reasoning models (OpenAI o3, DeepSeek-R1), and inference-time scaling.

Hard, verifiable state layer for AI agents

CI-grade evidence and safety tools for AI agents, MCP servers, and upstream contributions

Grafana AI observability plugin for Hermes Agent

160,000 deductively-derived JSON constraints that eliminate LLM persona drift. Measured: 0 breaks in a 120-turn pressure test (plain prompt: 8).

AI agent for X/Twitter with full read + write capabilities. Free reads via twitter-cli, writes via official X API MCP server.

Локальная мульти-агентная система на self-hosted LLM (Ollama). 3 канала (Telegram/консоль/MAX), 4 типа моделей (LLM/embed/vision/voice), мультимодальность (текст/PDF/фото/голос), роли Planner/Executor/Critic с рефлексией. Инструменты: почта, Яндекс.Диск, web, OCR, cron-планировщик. Семантическая память (sqlite-vec). Coverage 88%. Без облачных API.

Local-first terminal mission control for Claude Code, OpenAI Codex CLI, Gemini CLI and OpenCode. Run AI coding agents in parallel tmux sessions, manage direct SSH connections, monitor usage and continue securely from phone via Tailscale. No ComandOS cloud, hosted database, telemetry or SSH proxy.

Pre-deploy static scanner for risky AI agent capabilities. Local-only. No telemetry. No network calls during scan. Redacted output. MIT.

Local-first, human-in-the-loop personal AI employee. Watches email/LinkedIn, drafts actions with OpenCode, and waits for your approval via Telegram/Discord, before anything real happens.

Production-grade skills, custom MCP servers, and real projects built with Grok Build 0.1 (grok-build-0.1). The definitive toolkit + living showcase for xAI's agentic coding model. Sole author & maintainer: Cobus Greyling

Portable Goal-Based SDLC for Claude Code — interactive goal-mode + autonomous park-and-continue loop-mode, installable into any repo as a plugin. Zero runtime deps.

Knowledge vaLidated Expert AI assistant for neuroscience

A collection of LangChain / LangGraph/DeepAgents examples from the official documentation

Local-first RAG platform for technical document libraries. Features FastAPI, ChromaDB, and active retrieval (FLARE) powered by Ollama.

The open framework for governed AI development: EU AI Act compliance, PII redaction, MCP proxy, prompt-injection firewall. Python + Rust, Apache 2.0

A reproducible benchmark for studying when prompt optimization improves multi-agent LLM systems.

Learn from any GitHub repo systematically with AI agents guiding you through architecture, flow, patterns & takeaways

AI research agents + scientific image viewer for bioimaging & neuroimaging — LangGraph agents, GPU inference, and native in-browser OME-Zarr / TIFF / NIfTI / HDF5 viewers.

An embodied agent framework for orchestrating long-horizon, multi-module tasks in real-world robotics.

A local AI-powered OS assistant for Windows that organizes files, monitors system pressure, suggests cleanup actions, and stays privacy-friendly by running with local tooling such as Ollama.

Agent & worker factory (in construction) — self-hostable hub; tools plug in over NATS, CLI, skills, API

Minimalist coding agent harness with a Textual TUI and headless CLI. <1k-token system prompt, 18+ LLM providers, AGENTS.md + skills context, session tree, prompt/auto permissions.

Claude Code skill that invisibly rates every external resource (WebFetch / WebSearch / MCP / curl) via PostToolUse hooks. Two rater backends: Anthropic API key or claude.ai subscription via 'claude -p'.

Command Autopilot — use 100% of Claude Code while memorizing zero commands. Silent automation, right-moment suggestions with benefits, self-evolving to your habits.

MCP server giving Claude AI access to 20+ NYC public-record datasets across 9 city agencies via 18 tools — NYC property due diligence in plain English: violations, liens, sales, ownership, permits, rent stabilization, and more.

Wisdom before intelligence. A provenance-aware reasoning layer that abstains instead of fabricating — stops LLMs inventing attributions and merging distinct traditions. Measured: 0% fabrication on unknown-answer traps (vs 17–25% raw); −12.5pt hallucinated attributions at 0% false-positive cost. An AGI-candidate proof package, not proven AGI.

Agent-first GraphRAG workflow that turns local documents into reusable knowledge graphs. Drop files in, let skills extract entities and relationships, then query a persistent graph artifact locally—no cloud or pipeline setup required also the graph can be exported to neo4j and qdrant for deployment

[EMNLP 2025] M-LongDoc: A Benchmark For Multimodal Super-Long Document Understanding And A Retrieval-Aware Tuning Framework

抄袭者 Skill - 把开源项目抄过来变成你自己的 | 兼容 Claude Code / Codex / OpenClaw / Hermes Agent | AI Agent Skill that clones, rebrands, and customizes open source projects