Custom agents

Coding Agent Runtime & Evaluation Harness for controlled repository-level software repair

Dissenyador visual de workflows i graphs d’IA agentiva · Diseñador visual de workflows y graphs de IA agéntica · Visual designer for agentic AI workflows and graphs

A local terminal control plane for Codex, Claude Code, and pi sessions—discover, monitor, resume, and work beside persistent workspace shells.

Retrieve by relevance, not resemblance: filter an AI assistant's memories with TypeSafe's Jev

A pure-Python system that automatically discovers, verifies, and applies relevant historical requirements for coding agents — no embeddings, no vector database, no LLM calls. Companion code for the Towards Data Science article on intent continuity.

Local-first desktop app for audio AI: conversational agent + on-demand open-source model store.

Create an AI friend and build things together!

Fast dual-era framework-agnostic Model Context Protocol (MCP) runtime for Java and Kotlin. Streamable HTTP, native Netty transports, tools, tasks, resources, prompts, completions, skills over MCP, stateless and stateful with resumable session support.

The best news MCP server — 150+ verified articles/day with confidence scores, Ed25519 provenance, and Ethics Engine ratings. Free, no auth. Works with Claude, ChatGPT, Cursor, and any MCP client.

LLM agent 的有界自纠错长期记忆层：把连续、含矛盾的项目交互日志蒸馏成有界、可自更新的记忆库。Bounded, self-correcting long-term memory for LLM agents — candidate pool + value dynamics + tension adjudication + causal-replay evals.

A Model Context Protocol (MCP) server for integrating the Pinterest API with AI agents and LLMs.

Pay-per-call APIs and MCP services for AI agents: document OCR, document analysis, media transcription, browser sessions, and agent inboxes. USDC on Base over x402, no API keys, receipt-backed results.

OmO agent provider and workflow UI for Paseo - DAG graphs, live todo cards, approval popups, mobile-ready

The Turbo Haskell IDE

Takes for Mac (gettakes.app): a free, open-source Mac app for talking-head videos. A teleprompter that follows your voice, camera and screen recording, and your agent (Claude Code, Codex or pi) writes the script, cuts the edit and drafts the posts through the takes MCP server. Apple silicon, macOS 15+. Shared as is.

RL environments with verifiable rewards you can trust. Sandboxed coding and computer-use tasks, verifiers kept apart from the answer, a QA gate against reward hacking, and infra failures kept out of training. Works with verl, TRL, OpenRLHF and Axolotl.

Open-source platform for building and running apps, custom AI agents, and business automations on your own infrastructure. Includes reusable skills, external service integrations, per-project permissions, and Kubernetes hosting.

Use your OAuth subscriptions or API keys in any coding agent, Factory Droid, Cursor, and more. Tunnel Agent handles provider authentication and routes requests through CLIProxyAPI or 9Router.

A local AI agent built from scratch in Python with Ollama: conversation memory, system prompts, and multi-step tool calling.

Open-source, self-hosted platform for running a team of AI bots that use tools, keep long-term memory, and hand work to each other.

Argolink-focused coding-agent request and tool workflow for Claude Opus 5.

自己上传表情包，让 AI 看得懂、也会发 · Stickers your AI can read and send

ChatGPT, related application + prompt engineering list

Bring Claude (and any agent) into Overleaf. Claudeleaf is a reliable TypeScript bridge that drives the Overleaf web editor in real time — listing your projects, opening documents, reading them, and inserting/changing/deleting text — exactly as if a person were typing, so collaborators see the edits live. It ships as SDK, CLI, and MCP server.

Skills for managing an Amazon seller account via SP-API, through the Amazon Selling Partner Connector (MCP). Write actions are always human-in-the-loop (drafted for your approval).

See every Claude Code and Antigravity agent running on your machine — sessions, subagents, models and live status, inside VS Code.

Agent Skill that lets Claude Code, Codex, Cursor, Copilot and other AI agents set up and harden a Debian/Ubuntu VPS without locking you out: SSH, firewall, fail2ban, updates, Docker, TLS, backups, phone alerts. Benchmarked on 120 VM runs.

Text-to-CAD tools in Elixir with build123d, OpenSCAD, replicad, and smlibpy.

Private meta-search and answer engine in Rust: 17 engines, consensus ranking, full-page reading, and cited AI answers with any LLM. One binary, runs on your machine.

Interactive matte / roto tool for VFX – repairs damaged plate frames, generates alpha mattes for multiple objects (SAM3, CoTracker3, MatAnyone2, VideoMaMa, ViTMatte, LTX 2.5 Alpha Gen via ComfyUI), exports to Nuke and After Effects. REST API + CLI for AI agents. GPL-3.0.

Escouade, l'environnement de développement pensé pour les agents (ADE) : planifier, coder, relire et tester, livrer, avec des agents en parallèle. Windows, macOS et Linux, gratuit et open source.

A tiny autonomous whale-girl living on an idle Android phone. DeepSeek + AutoX.js + ntfy.

Unofficial read-only MCP server for Snappfood and SnappMarket: search dishes and groceries, compare true prices with delivery, read menus, reviews and flash deals from Claude or Cursor. اسنپ فود، اسنپ مارکت، سفارش غذا

Remote access to multiple devices through one MCP, with shared Recall memory, Workflow data, and centralized device status.

PTXBench: an open-source environment to evaluate GPU kernel languages for LLM agents

RF/microwave design automation + LLM orchestration: 16 EM/EDA engines behind one interface, 75 parameterized template families, 340 deterministic calculators, 87 calibration anchors; AI agents drive everything via MCP (157 tools) — physics numbers come only from deterministic kernels, never from the LLM. 28000+ unit tests, GPL-3.0

Forge Claude Code skills generate more reliable agents

面向地理学、遥感科学、土壤学、生态学的 Claude Agent Skill 四件套：研究论文写作（sci-paper-writing）、综述与 Meta 分析（geo-review-writing）、审稿意见回复（geo-reviewer-response）、个人写作风格档案（geo-writing-style）。覆盖定量反演、时空变化与观测对照三类论文，对标 RSE、ISPRS J、TGRS、Catena、Geoderma 等期刊。

51 Agent Skills for the Scavio web data API: one umbrella skill plus one per platform for Google, Amazon, TikTok, LinkedIn, YouTube, Instagram, Reddit and 40 more, as structured JSON

Enterprise-grade graph-powered code intelligence engine. Deep structural awareness for Laravel, PHP, Python, and JS. Powers AI agents with high-fidelity architectural context via MCP & Neo4j.

Web development framework with 17 commands and 5 agents for Claude Code - React, Next.js, Vue, TypeScript

Claude Code plugin that offloads large outputs to filesystem and retrieves when required.

tiny_fnc_engine is a minimal python library that provides a flexible engine for calling functions extracted from a LLM.

Pocket-sized Ruby AI agent framework / LLM assistant with multi-LLM support

Chat bot based on Bing web ai bot.

This is a Node.js console app containing Whatsapp handler using reactive methods to response messages as chatbot. Simple but useful project to start developing a complex NON-OFFICIAL Whatsapp chatbots.

A sophisticated AI-powered workspace built on top of Claude Code SDK

Multi-agent AI system for automated code review using Google's ADK - Kaggle Agents Intensive Capstone Project 2025

a tribute to multi-agent pathfinding composed of simple mechanics, borrowed art, and questionable music taste

[ACL 2026]From Experience to Skill: Multi-Agent Generative Engine Optimization via Reusable Strategy Learning

Fully self hosted chatgpt agent alternative

axe - a precision agentic coder. large codebases. zero bloat. terminal-native. precise retrieval. powerful inference.

Curated OpenClaw resource library with setup guides, personas, memory systems, security hardening, MCP integrations, communities, and more. 190+ links.

Zero-dep agent workflows with structured approval gates and adversarial review.

Build AI-powered engineering tools in simple YAML

NGOAgent-SystemAgent

Claude Desktop

It is a seq2seq encoder decoder chatbot using keras and with attention

Source code for our EMNLP19 paper "Retrieval-guided Dialogue Response Generation via a Matching-to-Generation Framework"

Useful resources for building, improving, and deploying contextual chatbots and assistants with Rasa

A collection of Discord programs to run autonomously

Airbnb Messaging Bot (TOBOT) - AI assistant for Airbnb Hosts

:evergreen_tree: A Discord <-> Terraria bridge plugin for TShock

Scientific Document Insight Q/A

This is a Next js project that implements a conversational AI Agents using ElevenLabs' SDK. The application features a voice assistant interface that allows users to interact with the AI through voice commands.

Sample to show using Azure Functions to run action functions from Azure AI Services

An implementation of tic-tac-toe in C, featuring AI agents

simple ai agent for wezterm

An example Discord chatbot built on the Letta API, which uses a stateful agent (agent with memory) under the hood.

A framework for building and testing custom autonomous agents.

AI agent skills for building, operating and troubleshooting Apache Kafka applications. Topic audit, consumer lag, schema review, security, connectors and DLQ

An AI legal workflow running on Claude Code. Four legal agents — each with their own jurisdiction, knowledge base, and MCP tools — collaborate to produce legal opinions with full audit trails.

AI Trading Agent with On-Chain Identity (ERC-8004) — 82.2% OOS Win Rate, Walk-Forward Validated, Kraken CLI Paper Trading

Fastest way to create personalized AI agents

The Hierarchical Intrinsically Motivated Agent (HIMA) is an algorithm that is intended to exhibit an adaptive goal-directed behavior using neurophysiological models of the neocortex, basal ganglia, and thalamus.

The control layer for AI Agents

AI agent for universal behavior analysis (BehaveAgent v1.0).

Openclaw 🦞 AI Polymarket Trading Bot — TypeScript bot built with Openclaw AI agent. Predicts crypto price on Polymarket 5m BTC Up/Down markets and tracks results (paper or live) and automatically executes trading according to the prediction result.

Provides summarised output from various actions that could otherwise eat up tokens and cause crashes for AI agents

The front door to AISquare open source. Governance, debugging, and audit trails for AI agents — because "it just works" isn't an answer your compliance team accepts.

👾 AI agents that build and run companies with you.

The official CLI for TCL Smart Home — manage smart home devices from the terminal. Built for humans and AI Agents.

Identity preservation protocol for AI agents. Build walls around who you are.

Unified CLI plugin manager for AI agents and developers - discover, install, and orchestrate thousands of command-line tools

Precisely define coding agents with different LLM parameters.

A simple example of how to implement vector based DDPG for MARL tasks using PyTorch and a ML-Agents environment.

Code for "Coordinated Exploration via Intrinsic Rewards for Multi-Agent Reinforcement Learning"

DocHarbor: portable multi-agent document retrieval and evidence workflow

Automatic prompt optimization framework for multi-step agent tasks.

Paper Collection of Reinforcement Learning Exploration covers Exploration of Muti-Arm-Bandit, Reinforcement Learning and Multi-agent Reinforcement Learning.

code for the accepted IEEE TITS paper: "Heuristic-based Multi-agent Deep Reinforcement Learning Approach for Coordinating Connected and Automated Vehicles at Non-signalized Intersection"

一个为 Claude Code 设计的简化版多智能体研究系统，通过并行子代理执行基于 Web 的综合性研究。 概述 本系统使用三种专门的代理类型： 研究主导代理（Research Lead Agent）：统筹研究、规划任务、委派给子代理、综合发现 研究子代理（Research Subagents）：使用 Web 工具并行执行专注的研究任务 引用代理（Citations Agent）：为研究报告添加适当的引用

Improve the output of coding agents like Claude Code and Github Copilot by providing the right context.

a peak-calling and differential analysis tool for replicated ChIP-Seq data

GPT-4 Function Calling Example in PHP

A demo of integrating React, Vite and Nitro for AgentConf 2023.

Build a Dialogflow chatbot frontend using Django framework and deploy it on App Engine.

A data access control framework for Open Policy Agent

This project designs an Android remote control framework based on customized Accessibility Services and HTTP/HTTPS communication, enabling remote device manipulation within a local area network. Furthermore, an AutoGLM-based intelligent agent is integrated to provide a cross-platform Mac UI for high-level semantic control of mobile devices.