Custom agents

The local catalog of your projects — intelligent, always learning, for you and your agents

A physics-aware multi-agent LLM framework for accelerating the material science discovery and optimization

A grumpy Gibson construct living in an office router. It only knows what the router can see.

A modular local AI agent built with Python and Ollama, featuring tool calling, memory, web search, document processing, embeddings, and RAG.

High-performance coding agent toolkit MCP server — IDE-grade code intelligence for LLMs

Plannable is a Spec-Driven Development planning CLI for AI coding agents. It creates a MASTER_PLAN.md plus compressed Sub-Plans phase files, so Codex, Claude Code, Cursor, and terminal agents can build software progressively with scenario hints, validation, and evidence-based completion.

A modern, sovereign AI backend for products and enterprise deployments.

RAG in your browser — query YouTube videos, podcasts & PDFs; cited answers deep-link to the exact video moment. Next.js + Gemini.

Production-ready Claude Code configuration for backend/full-stack developers. 9 rules, 8 commands, 5 agents, 13 skills, hooks, and MCP servers for Java/Spring Boot, Python/FastAPI, JS/React, PostgreSQL, MongoDB, Redis, Kafka, Docker, K8s, and AI/ML.

Unified multi-agent orchestration layer for Claude Code with parallel execution, session persistence, and natural language routing.

The official code base for paper AAAI2026 <MedLA: A Logic-Driven Multi-Agent Framework for Complex Medical Reasoning with Large Language Models>

An agent is a file. Jet is a dynamically typed, Ruby-flavoured language on the Erlang VM (BEAM) where object = actor = agent: one file declares a model, a role and what it exposes, and serves it over ACP or MCP — supervised, hot-loaded and distributed by OTP.

Reproducible benchmark and reference pipeline for reranked long-term memory retrieval

Benchmark Test-Time Scaling of General LLM Agents

AgentRAN: Multi-Agent Framework for AI-RAN Control

Interactive UI for AI agents — turn structured questions into browser forms.

The same HTTP API in 10 languages at 3 sizes, 30 implementations. Token cost for LLM agents, plus cold start, latency, throughput, CPU and memory. All measured, all verified over live HTTP.

Agent skill for authoring PR-ready RSSHub routes, distilled from 2000+ PR reviews

Connect and benchmark finance agent systems in one open-source platform, then trade on them. TradingAgents and AI Hedge Fund today, your own system through one adapter: a live front end for every agent, a database of every run, LLM and framework benchmarks, backtests, Interactive Brokers paper trading.

Self-hostable, multi-tenant AI agent service with sandboxed code execution（可自托管的多租户 AI Agent 服务，支持沙盒隔离执行）

多Agent金融研究报告自动生成系统 | Python/Java/Go三语言实现 | Pipeline+Fan-out并行架构 | 面试全套材料

React debugging in Chrome DevTools: a 5-view panel (Dashboard/Profiler/State/Effects/Settings) with render+memory profiling, issue detectors, a Redux inspector, AI analysis, and an MCP server for coding agents.

Concise, curated resources for working with the Clojure Programming and LLM base coding agents

Turn your AI agent into a hacker by plugging in this MCP

Mercury CLI: multiple AI coding sessions in one terminal. Mix and match provider and locally hosted models, spawn cross-company agents & workflows, and use built-in debugging, IDE built-in, screen control, and voice dictation.

Claude Code / Codex CLI skill that scaffolds a runnable UVM testbench (APB / AHB-Lite / AXI4-Lite + generic-bus fallback) from an IP spec and register table, with a UVM-sequence flow for DV and a tb_api:: task BFM for DE.

Open-source desktop AI trading assistant for market research: chat beside live charts, price alerts, local backtests, bring your own model.

Scene-Designer: Learning Design Intelligence for Agentic 3D Scene Generation

A sandbox service for running untrusted, agent-authored code that reports the isolation boundary each run executed behind. Lessons, no clone needed: https://plimsollmark.github.io/plimsoll/trainers/

MOVED to dsh-github-bot — https://github.com/PerryLink/dsh-github-bot. This repository is a pointer/backup; the old npm package @perrylink/dsh-github is deprecated.

J.A.R.V.I.S. voice assistant with an Iron Man holographic UI, driven by Claude Code. Built on adewaskar/jarvis, with Windows support and a loopback-only bridge.

AI 漫剧/短剧连载创作引擎（Agent Skill）| 一句话到剧本+分镜+配音+AI提示词的工业化流水线 | 七个记忆文件+22项跨文件一致性检查，解决长线连载人设崩坏与伏笔丢失 | Agent-agnostic (Claude Code / Codex / Copilot / Hermes) · MIT-0

Cross-device DeepSeek Harness session sync: a dedicated git mirror with append-only keep-both conflict resolution (fork files, never silently overwritten), /sync command and sync_* tools

基于 Price、Wieman 等人 29 Decisions 研究的科研决策 Skill：结合已有结果、可追溯方法与领域资料，解释下一步为什么值得做。支持多种 Agent Skills 宿主。Research decision support for STEM.

MCP server for real-time equity research: real-time market events, 13F ownership, earnings transcripts, fundamentals, estimates & insider filings on 2,900+ companies for Claude and other AI agents.

Model-agnostic AI agent skill (open Agent Skills format) that reviews draft ITU-T Recommendations against the ITU-T Author's Guide, presentation rules and quality checklist. Works with any LLM.

[ICLR 2026] Official implementation of "ChatInject: Abusing Chat Templates for Prompt Injection in LLM Agents"

Structured skill packs for AI agents (Claude Code, Codex, Cursor) to operate the Horizon AI toolchain — HBDK compilation, HMCT/Plugin quantization, UCP on-board inference, and LLM compression.

2026 papers on long-horizon tasks: LLM agents, VLM planning, VLA manipulation

An evidence-grounded LLM agent for diagnosing PyTorch training failures and verifying proposed repairs through bounded retraining.

A curated, categorized reference of AI-powered coding tools. Covers full IDEs, editor extensions, terminal agents, autonomous agents, browser-based builders, and code review platforms.

Taller práctico de Google Agent Development Kit (ADK) con Gemini, MCP y multi-agentes

AI multi-agent harness that plans, designs, and produces online courses (slides + notes + TTS audio + quiz + player) from a topic using ADDIE + Bloom's Taxonomy.

Risk-aware multi-agent deep reinforcement learning for packet routing in ultra-dense LEO satellite networks

Persistent multi-view corpus navigation for evidence-efficient agentic search

Vanilla RAG baseline for the Agent Memory Leaderboard

A privacy-first, local-first AI agent framework written in Rust

Kitopia -属于你的新一代桌面全能效率神器 Your Next-Gen All-in-One Desktop Productivity Suite

CoEvolve: Training LLM Agents via Agent-Data Mutual Evolution

Official implementation of the paper "NLI4VolVis: Natural Language Interaction for Volume Visualization via LLM Multi-Agents and Editable 3D Gaussian Splatting"

MCP server for Ruby runtime debugging — connects LLM agents to live Ruby/Rails processes via the debug gem.

Autonomous and goal-seeking coding agents. 💫

A Strong Agentic Framework for Model Adaptation and Auto-Evaluation

Frappe, ERPNext, and Frappe ecosystem agent plugin for Codex, Claude Code, Cursor, Copilot, Helpdesk, CRM, LMS, Drive, HRMS, Insights, Builder, and Payments.

A curated list of open-source medical and healthcare AI projects - LLMs, imaging, multi-agent systems, clinical software

面向建筑行业知识工作从业者与 AI 研发团队的 Skills、Workflow 与多 Agent 工具包 / Building-industry AI agent skills for BIM, IFC, RAG, GraphRAG, project evidence work, code review, and runtime governance.

Biến Grok Bot thành lớp điều phối, DeepSeek Harness làm worker giá rẻ — tiết kiệm quota, chạy liên tục

Multi-step automations for Hermes Agent, authored as a graph and running on native Cron, Kanban, and Profiles.

One framework to ship on every chat interface. MCP + CLI, WhatsApp, iMessage agents, ChatGPT, Claude Apps, and Muse Connectors

MoUI — a multi-platform declarative GUI framework and AI-agent workstation for MoonBit. 多平台 MoonBit 声明式 GUI 框架：显式运行时管线 + 可插拔渲染器。

这是一套围绕 Stanford CS224N: Natural Language Processing with Deep Learning 整理的系统课程笔记。仓库以课程 19 节 Lecture 为主线，从 Word Vectors、RNN、Attention、Transformer 一路延伸到 Pretraining、Post-training、PEFT、RAG、Agents、Reasoning、Multilinguality、Interpretability 与 Multimodality。

158 expert AI skills for ROS 2, Nav2, perception, SLAM, localization, URDF, Gazebo, and robotics development with GitHub Copilot.

Evidence-backed retrospectives for AI coding-agent histories. Offline indexing, scoped workflow improvements, and bilingual documentation.

Unreal Engine 5.8 without the editor: TypeScript, declarative React scenes, Vite hot reload, built for AI agents.

Lightweight AI chat interface built with React and Golang.

NEXO runtime core for NEXO Desktop: local memory, automation, MCP tools and update-managed runtime.

PIT 约束下的 AI 投研 Agent：LangGraph 多 Agent 编排（并行取证 - 辩论 - 综合 - 代码化风控 - HITL）+ PIT RAG + 回测/模拟盘。

Simple issue triage without GitHub Agentic Workflows. One small Ruby script, cheap Copilot prompts, cached replies. Stop maintaining your issue bot.

哪吒网络安全是一款高性能终端 UI (TUI) 应用，专为从事红队演练、渗透测试和漏洞研究的网络安全专业人员打造。本应用作为智能指挥控制界面，利用 DeepSeek 大语言模型提供实时流式对话、基于函数调用 (Function Calling) 的自动化工具编排，以及多智能体会话管理——这一切均在轻量级终端环境中完成。 核心设计哲学：采用 Elm Architecture (Model-Update-View) 模式，纯函数状态转换，通过 tokio::sync::mpsc 通道传递事件，遵循零成本抽象的 Rust 惯用模式。

Beautiful docs for humans. Native tools for agents. An open-source Astro framework for documentation you own end to end: Markdown in, static site with search, an API reference and an MCP server out.

模仿海马体效果的DSH 的长期记忆插件，为单个工作区安装后可实现工作区内的跨会话窗口记忆

基于 Pi 的独立桌面编程客户端：在一个窗口中管理项目、模型对话、文件、Git 与终端，支持 Skills、Plugins、工作区信任和会话权限设置。

一个基于 twikit 实现的 X (Twitter) MCP 服务端，轻量级工具。支持AI调用X平台功能，包含发推、回复评论、读取推文回复、私信收发、抓取用户主页详细信息、查看推文互动数据、推文搜索、点赞、转推、删推等多项工具。适用于 Claude Desktop、Cursor 等支持 MCP 协议的客户端。

Stop debugging the same errors twice. CoMemor equips your AI coding agents with repository-scoped persistent memory, turning past command failures and fixes into instant solutions. Whether you're building a personal side project or scaling complex enterprise codebases, give CoMemor a try today and experience seamless, self-learning AI workflows!

网文平台资源配置决策助手：GraphRAG 两阶段检索 + ReAct Agent + 规则引擎兜底，方案合法率 60%→100%

Unofficial read-only MCP server for Iranian job sites (استخدام) IranTalent, Jobinja and Jobvision: job search, companies, reviews, interview reports and pay side by side. For Claude, Cursor, Copilot and AI agents.

A repository with academic works about Kernel generation with LLM Agent

Meta Ads Skill is designed to give LLMs and AI agents expert-level capabilities to orchestrate the Meta Ads MCP Server.

Meet Klex Bots - Your own team of digital coworkers. Fully autonomous agents that work like your team does. Self-hostable and model-agnostic.

写给算法工程师的大模型课程：34 课，每课从一个问题出发，用具体数字把机制从头算一遍

A proactive AI agent for secure, traceable, human-in-the-loop task execution over long-running workflows.

Seed 是一个专为游戏研发设计的 Claude Code 插件。 一条命令描述任务，Seed 自动分析类型和领域，从五个专化 Agent 中组出最合适的组合，通过 Claude Code 原生 Team 机制启动协作。实现、调查、修复、审查、Unity Editor 操作，每种任务都有对应配置。 内置项目记忆系统，跨 session 自动保持上下文，不需要每次重新交代背景。

AI security agent for the Python supply chain: scans packages, generates exploits, and validates them in Docker, autonomously.

Polarity.ai: The AI that refuses to be neutral. It either defends your worst ideas or attacks your objective truths.

Agentic framework for physics

Multi-platform ACP client (VS Code + CLI) for connecting AI coding agents — fork of vscode-acp with Windows fixes & improvements.

SWE-bench for your codebase — mine your merged PRs into local, contamination-free coding-agent benchmarks. Adapters: claude-code, aider (Opus 4.7 / GPT-5.5 / Sonnet 4.6 / Gemini 3.1 Pro).

Distributed multi-agent research engine with dynamic strategy planning, durable stream-based coordination, and controlled synthesis

AI-Powered Multi-Agent Crypto Investment Analysis Platform

Build your own Chat GPT with python. Don't forget to star 🌟 this repository.

Anil AI is an innovative chat application that harnesses the robust capabilities of Generative AI, utilizing the advanced LLAMA 2 model for inference.

🧠 Personal AI Gateway — Single-file Python AI agent with multi-LLM, tools, vision, TTS, encrypted vault. Your own ChatGPT on localhost.

Multi-agent version of highway_env simulator

Implementation of the paper "Reciprocal Velocity Obstacles for real-time multi-agent navigation" in ROS using rospy in Turtlesim

Djinn-Agent: A lightweight CLI tool for seamless interaction with Claude's advanced computer-use capabilities, automating complex tasks from the terminal.

NextJS,React,Lama,VAPI,Twillio,Retell,Telegram,Sales

Your coding agent forgets everything between sessions. Lore fixes that.

CLI Coding Agent for professional Agentic Engineers focusing on output quality. Self-review, supervision, senior-style, efficient and fast.

ApexPanda 官方仓库 · 仅供合法合规使用，禁止用于任何违法用途

Enhanced Voice Assistant, Works with both terminal and API, Multimodal, Multilingual, Modular design. Supports Voice ID, Face recognition, and configurable tools. Built-in Chatgpt, Claude, Deepseek, Gemini, Grok, and Ollama. Explore the possibilities of Human-AI interactions