Custom agents

ManiSkill-Learn is a framework for training agents on SAPIEN Open-Source Manipulation Skill Challenge (ManiSkill Challenge), a physics-rich manipulation skill benchmark with large-scale demonstrations.

[ICLR 2024] Trajectory-as-Exemplar Prompting with Memory for Computer Control

Compiler-driven MCP framework for Kotlin Multiplatform

Design and implement agents, bot and chatbots in Python

Specs, tracking, and coordinated development for AI coding. Works with Cursor, Claude Code, Replit Agent, and any AI coding assistant.

Developed an AI-powered supermarket analytics system using LSTM, K-Means, and Apriori to predict sales, segment customers, recommend products, and optimize inventory through an agent-based architecture with a Streamlit dashboard.

CodeNav is an LLM agent that navigates and leverages previously unseen code repositories to solve user queries.

Unofficial PyBrain extension for multi-agent reinforcement learning in general sum stochastic games.

Multi-agent coordination for Pi - presence, messaging, file reservations

Aries agent test framework, with agent backchannel support

Privacy-first agentic framework with powerful reasoning & task automation capabilities. Natively distributed and fully ISO 27XXX compliant.

A self-contained multimodal AI agents lab built using MongoDB, Gemini and LangGraph.

A seed for growing a personal AI that improves through use and continues through failures.

A suite of tools including a framework for creating Java Agents, for aspect-oriented tooling for distributed systems.

Workshop "Build a Retail Copilot Code-First on Azure AI": attendee and presenter resources.

Compass is a context engine that builds a knowledge graph of your organization's metadata, capturing entities, relationships, and lineage across systems and time, making it discoverable and queryable for both humans and AI agents.

Slax Reader goes beyond saving content—it helps you learn. Save web pages, highlight, discuss, and use AI to clarify concepts. Follow creators and explore shared bookmarks for collective insights.

An AI chatbot plugin for Obsidian using the Gemini API for note summarization, content generation, and more. Enhance your workflow with AI assistance like the Notion AI bot.

Give your AI agent personality, memory, and autonomy. Ready-to-use workspace template for OpenClaw.

Clawdbot skills for agentic coding workflows - ACFS stack, cloud CLIs, and dev tools

Extension to the Mesa repository to provide with the ability to plug LLM directly into your agents for more complex reasoning and inter-agent communication ! Make your own reasoning/memory/communication modules and plug them into the existing code or use the pre-built ones for fast building.⁦

Give your AI team a visible office. Pixel-art RPG workspace for multi-agent collaboration.

An AI agent to automate dating using hinge

AI-powered content pipeline: trend discovery → research → social publishing. Built with Claude Agent SDK + MCP.

Self-hosted alert platform for ops teams. Telegram Bot API compatible. ACK, push, team chat.

Run AI coding agents in Podman or Docker

A VS Code extension that embeds an AI coding agent (Claude and GPT and other models) in your editor, powered by the Pi agent runtime.

A skill for coding agents that build, review, and refactor Web UI that should be easier for humans, screen readers, browser automation, Playwright tests, and AI agents to understand and operate.

Open-source self-hosted web tool for evaluating Agent Skills with rubric scores, Deep Review, and improvement suggestions.

AWS Certified AI Practitioner (AIF-C01) exam preparation

项目描述：构建⼀个能够⾃动撰写多模态呈现、具备专业性和深度、数据融合与事实溯源、规范 有逻辑的各类⾦融研报的智能 Agent 系统。 主要负责：根据赛题思路，对目标公司生成金融研报，通过 LLM 获取目标公司的竞争对手，用 akshare 获取数 据源的的三大报表数据，通过 duckduckgo 获取公司信息、行业信息、股份信息等，通过设计数据分析师智能 体，包含三个动作代码生成和执行、收集图片、完成动作，用 pyplot 生成金融报表。最后通过 word 形式文件 展示多模态研报。包括目标公司财务状况、关键财富指标分析、财务问题等，在天池大赛 A 榜单获得前百。

A coding agent-first CLI for interacting with LangSmith.

from vibe coding to agentic engineering - practice makes gemini perfect

ChatAllAI是一个基于Electron + Vue3 + TypeScript构建的桌面应用程序，旨在解决用户需要同时与多个AI模型进行对话的痛点。通过单一界面，用户可以同时与 Gemini、ChatGPT、Grok、MiroThinker、DeepSeek、豆包、Qwen、Kimi、Copilot、GLM、元宝等主流AI模型进行对话，实现高效的多模型对比和协作。

Plain-Markdown persistent memory for AI coding agents. Long-term facts, daily logs, scratchpad, and semantic search — works with pi, Claude Code, and any LLM agent.

My subagents repo

Open-source framework for agentic quantitative finance research.

[ICML 26] An evaluation framework assessing long-context retention and long-horizon memory performance for agentic applications (AMA-bench).

Modular MCP server for OPNsense firewall management - 88 tools providing access to 2000+ methods through AI assistants

One contract for execution governance across every stack, every team, and every AI agent.

Agent Skill for generating AGENTS.md files following the agents.md convention | Claude Code compatible

AI agentic skills framework for spec-driven development, built on the organizational model of mafia.

Claude Code skills and sub-agents for .NET Developers

PM Copilot: 65+ framework-grounded PM skills across 12 domains, 17 command-based workflows, and 8 specialized sub-agents. Built by Product Faculty.

Projx — A CLI tool for scaffolding production-grade projects. Pick your stack (FastAPI, Fastify, React, Flutter), get a fully wired template with auth, database, CI/CD, and E2E tests. Update anytime when templates improve.

The fastest way to build real-time Conversational AI and Voice Agents with Claude Code, Cursor, Copilot, Windsurf, or any AI coding assistant — powered by Agora.

Add CometChat chat & messaging and voice & video calls to any React, Next.js, React Native, Angular, Android, iOS, or Flutter project through your AI coding agent. Works with Claude Code, Cursor, Codex, VS Code Copilot, Windsurf, Cline, Kiro, and 50+ more agents.

🧠 Multi-Agent AI System running 100% offline using Google Gemma-4 models. 5 specialized AI agents collaborate in infinite loops for brainstorming. Features: Dynamic VRAM swapping, 4-bit NF4 quantization, WhatsApp-style RTL UI, live token streaming, AI-powered agent customization. Built with PySide6 + PyTorch + Transformers.

The first open-source @base agent.

Valorant 2026 Ultimate Agent Toolkit – Optimizer, Xray & Customization Suite

Loop engineering for Claude Code and other agent CLIs: figure the problem out first, define what done means, then execute and verify every criterion independently.

An ntfy MCP server for sending/fetching ntfy notifications to self-hosted or ANY ntfy.sh server from AI Agents 📤 (supports secure token auth & more - use with npx or docker!)

Self-improving AI agent companies — Paperclip V2 fork with persistent memory, MCPs, skills management, and KPI tracking. Built by HarborSEO (harborseo.ai)

Skill de Claude Code para crear automatizaciones (agentes de IA) en Cloudflare desde cero, en español sencillo. Por Horizontes IA.

Local-first flight recorder for coding agents : replay every run as a live session map, score the context bill, and write the fix back into AGENTS.md — no API key, one npx command.

Content Writing Agent. Simple & Extendable. Use with any model so your content doesn't sound like Claude.

Stealth headless browser engine in Rust — native BoringSSL TLS/JA4 fingerprint, real JS via deno_core, no Chromium, no CDP. Python + MCP bindings. A camoufox alternative.

Pilot coding agents (Claude Code, Codex) and monitor their telemetry from one cockpit.

Principal-PM-level product-teardown skill — reverse-engineer any product as a system and render a bilingual (EN + ZH) HTML report with a product-screenshot gallery.

Give your AI agents grounded, reusable career context.

Born for agent systems

ThirdSpace Knowledge OS — 知识库操作系统模板。Agent-native，零配置，任意 AI Agent 均可驱动。

CUA-Gym-Hub: mock web apps as reproducible RL training environments for computer-use agents

AI coding agent tools to build connectors using Fivetran Connector SDK

用于让 agent 在大众点评自动取号,定时取号,符合条件后触发取号,以及定时获取排号情况等,全流程自动化脚本控制,无需ai手动执行

Claude Code plugin that orchestrates GPT Codex agents for implementation, monitoring, and independent reviews.

一个生成 「影视飓风(Mediastorm)」风格中文口播稿 的 Claude Code / Agent Skill。

An AI driven debugging assistant specially designed for x32/x64dbg, it helps to double the efficiency of reverse engineering, vulnerability mining, and malware analysis, making it your intelligent copilot.

Open-source agent sandbox infrastructure for parallel coding agents: isolated workspaces, MCP/CLI control, observability, and atomic publication.

🦀 Decoder-only LLM built from scratch in pure Rust using Candle — no Python, no PyTorch. Gated DeltaNet + sparse attention, fine-grained MoE, native video/document understanding, long-horizon tool agents, quantization-aware training. Scales: Tiny (25M) to Large (1.3B).

Autonomous LLM-driven Among Us simulation with memory, social reasoning, pathfinding, sabotage, emergent behavior, and real-time React/PixiJS visualization.

Official implement on 'What and When to Distill: Selective Hindsight Distillation for Multi-Turn Agents'

This project implements a supervised deep learning system that learns to autonomously control a vehicle in the Countryside level of Hill Climb Racing. Using computer vision and a convolutional neural network (CNN), the model observes raw gameplay frames and predicts the correct driving action in real time — mimicking human input.

MCP Deep Research Server using Gemini creating a Research AI Agent

AI Agent Memory System inspired by Neuroscience

ClawSync, OpenClaw for the cloud. Deploy an open source personal AI agent with chat UI, skills system, MCP support, and multi-model routing. Built on Convex.

An engaging 1-on-1 chat interface between you and your AI agent.

AxonFlow: Runtime control layer for production AI

CReact is how you write your workflows using JSX, build durable apps that you can sync with external state, build jobs, agents, infrastructure, anything you can declare you can build!

Chatbot for searching vocabulary on mainstream dictionaries

Yet Another LLaMA/ALPACA Discord Bot

FreeVA: Offline MLLM as Training-Free Video Assistant

A blazing-fast CLI tool to export a directory's structure and contents into a neatly formatted `.txt` or `.json` file.

Farmer Agent on BSC Chain: farm-themed AI agent dashboard, cloud console, airdrop festival, and deployment stack.

The SDK interface to Letta Code. Build deeply personalized agents with persistent memory that learn over time.

tRPC-Agent-Python provides an end-to-end foundation for agent building, orchestration, tool integration, session and long-term memory, service deployment, and observability, so you can ship reliable and extensible AI applications faster.

Next-generation UI automation harness powered by Python & Playwright. Chat with AI Agents to seamlessly generate, architect, and execute enterprise-grade UI test code.

Using tensorflow, this agent can autonomously train itself to play Out Run and potentially be modified to play other games or perform tasks other than gaming

My Claude Code Settings(agents/commands)

📱 Make apps accessible for AI agents

Claude forgets everything when the session ends. CodeSyncer makes it remember.

The GitHub Action for Promptfoo. Test your prompts, agents, and RAGs. AI Red teaming, pentesting, and vulnerability scanning for LLMs. Compare performance of GPT, Claude, Gemini, Llama, and more. Simple declarative configs with command line and CI/CD integration.

Prototype of Full Agentic Application Security Testing, FAAST = SAST + DAST + LLM agents

基于ChatGLM3-6b的智能对话系统，集成了RAG、知识图谱、Agent、多模态等技术来增强大模型的回复质量。

OS-Harm: A Benchmark for Measuring Safety of Computer Use Agents [NeurIPS 2025 Spotlight]

Enables developers to build, deploy, and manage multiple specialized agents that work together to tackle complex workflows. Each agent focuses on a specific task, breaks down processes into manageable steps, assigns tasks to the appropriate agents, and ensures accurate, and reliable outcomes.

A (not too) Evil Chatbot 👹

Use Multi-agent Twin Delayed Deep Deterministic Policy Gradient(TD3) algorithm to find reasonable paths for ships