Custom agents

Per-step LLM routing benchmark with 970 static labels, live SWE-bench evaluation, an open data pipeline, and a public leaderboard.

HTTP + WebSocket 反向代理：把局域网端口转发到本地 DSH 服务

小说创作模式：一个统筹队长统领全局，5 个专职子代理各司其职——架构世界、策划剧情、管理人物、执笔写文、质检复核——协同写作。

Agent-first multi-tenant Next.js template (rules, sub-agents, skills, Better Auth, Prisma)

Profession-specific AI plugins for Claude Cowork — pharmacists, PTs, real estate agents, and more

AI-assisted development toolkit for Magento 2 and Mage-OS. An MCP (Model Context Protocol) server that enables AI coding agents to interact with Magento and Mage-OS installations through introspection tools, code generation capabilities, and comprehensive Magento/Mage-OS knowledge.

Open-source agentic coding workspace for persistent agents, parallel sessions, and inspectable context across projects.

Type-aware, evidence-backed delivery contract for AI coding harnesses. Different agents in. One predictable human handoff out.

Prepare any codebase for AI. Wires OpenCode, OpenSpec, codegraph, and agentmemory into a multi-agent development workflow powered by native parallel subagents.

Autoregressive Thought Generation Agent Framework

VeRO is an evaluation harness for using coding agents to optimize LLM-based agents and workflows. It treats agent code as a versioned artifact — making changes, evaluating results, and hill-climbing toward better performance using git version control.

Claude Code Agent Skills

XR Python toolkit for humans and AI agents

A TUI agent orchestrator for power users.

Visual AI workflow orchestrator for DeepSeek Harness (dsh): multi-agent DAGs, live execution, recovery, and Feishu integration.

The first Agent Skill a beginner installs. Bootstraps Claude Code, Cursor, and other agents, then routes every task to the current best skill for that job.

Terminal coding agent with browser UI, MCP, Obsidian memory, and live office floor. Any model via OpenRouter. MIT.

Claude Code production-readiness audit plugin with 8 specialist lenses in 3 waves for security, reliability, tests, DevOps, live-app QA, and launch blockers.

The reliability & architecture analyzer for LLM and agentic systems. By ActaClad.

SteamPipeGUI for macOS: build and upload Unity games to Steam from the Unity Editor on Mac, with no steamcmd commands or VDF scripts to write. The macOS answer to Valve's Windows-only SteamPipeGUI: one-click Build and Upload, Steamworks SDK integration, multi-depot support, and a headless batch mode for CI and AI agents.

A terminal board for the conversations your coding agents leave behind — Claude Code and five more — grouped by what they are about, not by the folder they ran in. Resume any of them in place.

AtomMem: a long-term memory system for personalized LLM agents.

Sea safety intelligence dashboard — AI-powered dive condition forecasting for Dauin & Apo Island, Philippines. PyTorch LSTM forecaster with XGBoost baseline and LLM agent, pluggable weather/marine/air-quality providers (Open-Meteo, Storm Glass, AQICN). React 18 + Vite frontend, Python backend, Docker.

Sanitized multi-agent control example with candidate pools, quorum, constrained review, checkpoints, and replay.

Onesist platform to enhance productivity for System Analyst deliverable Technical Requirement

Multi-language Active Inference framework combining ant-colony-optimization-inspired multi-agent simulation with variational message passing, implemented across 50 programming languages for research and benchmarking.

Harness-based Agentic RAG: search, understand, and answer questions over documents and code using an AI coding agent harness (Cursor, Claude Code, Codex, etc.) in a sandbox — no embeddings, no vector DB.

Local OCR, metadata automation and RAG document chat for Paperless-ngx using PaddleOCR and Ollama, designed for CPU-only self-hosted servers

Your agents, briefed. Not flooded. Self-hosted context compiler: git-backed team knowledge served to coding agents as token-budgeted bundles over MCP.

Search electronic parts, download datasheets, and extract structured data for LLM agents — Mouser & DigiKey CLI in Rust

Native macOS git worktree manager and GUI. See every worktree in one window and watch files change live while Claude Code, Codex and Cursor agents work in parallel.

Systematic software development framework for AI coding agents upgraded with TypeSafe Jev System One typed decisions

Let AI Agents control your real firefox browser: fill forms, authenticate, automate actions, web development; all via token-efficient, easy to use CLI. Claude plugin / Codex skill included. agent-browser, but for Firefox chads.

Agentic Development Environment (ADE) for the AI era

Shared memory for AI agents, no database, no server, just a GitHub repo. Memories that worked get stronger; memories that failed fade.

Terminal multiplexer for coding agents, built for working from your phone. Live status for every Claude Code, Codex, pi and opencode session, quota per account, dev-server previews and left-behind processes. Self-hosted over Tailscale, no account.

A platform creating fully autonomous Hermes agents for your team.

Multi-agent system for resolving Site Reliability Engineering tasks.

基于《崩坏：星穹铁道》角色「流萤」的桌面智能 AI 伴侣系统。 流萤以 Live2D 形态常驻桌面，流式 AI 对话、本地 ONNX 语义记忆、双引擎主动关怀、GPT-SoVITS 流萤原声 TTS、Agent 任务执行、MCP 开放工具链与 Skill 插件体系。

Chimera: Harnessing Multi-Agent LLMs for Automatic Insider Threat Simulation (NDSS 2026)

OpenTelemetry tracing for DeepSeek Harness (dsh): turns each agent turn into a GenAI span tree — steps, LLM calls with TTFT, tool executions, token usage — exported over standard OTLP to Jaeger, Grafana Tempo, SigNoz, Langfuse, or any compatible backend.

AI coding skill for generating and modifying Metin2 client UI code from screenshots, descriptions, or existing scripts. Works with Claude Code, Cursor, Codex, Windsurf, Cline, Copilot, and Gemini CLI.

Espressif 全生态统一 AI 技能。Unified AI Skill for the entire Espressif ecosystem.

用对话操作 Synthesizer V Studio / Instrument X 的 AI 助手（基于 DSH）：改音符、音高线、参数与织体

Vim style TUI with Repository-Aware, Self-Evolving Agent That Understands Codebase in Real Time powered by Hybrid Semantic + Full-Text Engine

一个本地可扩展个人 AI agent 工作台（a local-first, extensible personal AI agent workspace）

基于 DeepSeek Harness 的多数字生命系统 | Multi Digital Life: native agents, independent life workers, durable communication and owner-bound private services.

ChatGPT 交个AI虚拟朋友（CP）吧 👥

基于 LLM 的白盒漏洞挖掘 Agent。四条挖掘路径（启发式、快速扫描 Sink 回溯、历史漏洞绕过、无约束）+ Docker 靶场 / 局部 harness / 静态验证，支持互联网复现与攻击链串联。 LLM-based white-box vuln mining agent. Four paths (heuristic, fast-scan sink backtrace, historical bypass, unconstrained) + Docker lab / harness / static verify; Internet retest and attack chains.

Your terminal, now autonomous. Zero-dependency terminal AI agent.

Multi-backend web search CLI for agents — SearXNG, Ollama, DeepSeek

Agent-first CLI tools, UX patterns for LLM agents to prove mathematical results

ChatMux — provider-agnostic tmux chat web-terminal for coding agents (GJC/Codex/Claude/Cursor/OpenCode)

Inverted code review: a read-first IDE where you comment on the lines that matter and your coding agent (Claude Code, Codex, Cursor, Copilot) resolves them. Free and open source.

Allows to red-team your AI systems through adversarial probing. It is simple, effective, and requires minimal setup.

A production-ready handbook for Forward Deployed Engineers (FDEs). Covers enterprise landing zones, VPC Service Controls, data auditing, multi-agent orchestration, and LLM evaluation (中英双语).

An illustrated multi-agent village strategy game. Set objectives; residents gather, build, upgrade and recruit. Battle, inspect decisions and test AI agents. No API key required.

SearchEyes: Towards Frontier Multimodal Deep Search Intelligence via Search World Simulation. A typed knowledge graph unifies data synthesis (PKC), environment simulation, and RL training (HaPO) for multi-hop visual search agents. SearchEyes-27B achieves SOTA among open-source multimodal search agents on 6 benchmarks.

Coordinate coding agent teams from GitHub issues to reviewed pull requests, with a local dashboard for progress, decisions, and human oversight.

An easy-to-use macOS sandbox for Claude and other AI coding agents

An actively maintained fork of the modular ja-netfilter Java Agent framework.

One Power BI template for Microsoft Copilot credit consumption and cost - Cowork/Work IQ, Copilot Studio, GitHub Copilot and Azure AI Foundry.

Clone your customers from real sales-call transcripts and run them as a standing focus group — /focus-group Agent Skill for Claude Code, Codex, Cursor, and more

[ACM MM 2026]UrbanWorld2.0: A Multimodal Agent Framework for Reality-Aligned 3D World Generation at City-Scale

An automated, multi-turn LLM red-teaming pipeline that pits an Attacker model against a Target model to evaluate safety boundaries, featuring LLM-as-a-Judge scoring and an interactive HTML vulnerability dashboard.

HyperFrames is an open-source framework for turning HTML, CSS, media, and seekable animations into deterministic MP4 videos. Use it locally with the CLI, from AI coding agents with skills, or as the rendering core behind hosted authoring workflows.

DSH 插件：把 WorkBuddy 桌面 App 里的模型接入 DeepSeek Harness，零配置直接用。（原生嵌入"插件-插件配置页"）

Example Go adapter that feeds WordPress articles into Dify for an AI reading and RAG demo with Moonshot.

The free SEO Agent from the video: 7 Claude Code commands that take a business from keywords to a live, optimized, indexed website.

An iPhone assistant that answers with native SwiftUI pages

Enterprise low-code aPaaS platform with dynamic forms, BPM workflow, RBAC, multi-tenancy, notifications, agent workflows and IoT integration.

A tested harness for durable work across coding agents: CLAUDE.md/CHECKPOINT.md state and 17 skills, a Claude Code plugin with enforcing hooks, and a public benchmark of what it costs.

Agentic PR reviews for the IBM Carbon Design System, run by IBM Bob or Claude from a codified review prompt and grounded by Carbon MCP. Built on a reusable agent harness for IBM Cloud Code Engine: scheduled, parallel runs, a per-run spend cap, a web console, Elasticsearch telemetry and a Slack digest. A skill adapts it to your repo, job or agent.

Security logging for AI coding agents: point Claude Code, Codex or any agent at your repo and add security events in small, verified waves. 90 review items, 9 workflow prompts, MIT.

Run Claude Code and OpenCode on your Android phone, no computer needed. Or follow and steer agents on your computer: live sessions, tool approvals and diffs.

an ascii manta that guards the $GLYPH reserve

Open-source observability for AI applications: every LLM call, prompt, token and agent decision.

Offline voice-agent application layer: session core with byte-deterministic replay, spoken-LLM adapters (scripted/replay/loopback-mock), injection-safe function-call grammar, slot grounding, workflow engine with compensation and resume, permissioned tool sandbox, repair-dialog policies, task-success and latency evaluation. NumPy core, torch extra.

A local wiki your agents write for you. Nimbalyst Wiki plugin for Claude Code and Codex: capture decisions, track typed pages, read and edit in your browser.

System One Coding Agents: code-level answers at the same score as Claude Fable 5.1 on the same 74-point xz benchmark (69 / 74, 93.2 % each), at 1/20 of the cost and 3.5× faster. No Context Loss and Memory Limits on sessions since every answer starts from the same, updatable, SQLite index.

Real-time context layer that converts user activity into actionable signals for OpenClaw.

Drop-in pytest plugin for regression-testing AI agents — snapshot baselines, semantic comparison, mock LLMs

AI dev that actually gets things done

RAG Agent System - Spring AI + React + Multi-modal Document Parsing + Hybrid Vector Retrieval

Code graph tool for AI assistants - understand codebase structure, dependencies, and impact

Botium CLI - The Selenium for Chatbots

This is a twitch chatbot based in Llama2 tecnology

The official implement of paper 《DaMo: Data Mixing Optimizer in Fine-tuning Multimodal LLMs for Mobile Phone Agents》

A multimodal agent built with Python agents framework

多代理(Multi agent)强化学习Qlearning算法在多目标探测问题(任务分配+功率优化)中的应用

🦞 OpenClaw uses a multi-agent approach to simulate a complete IT team, automating the entire process from requirements to development, testing, and delivery. OpenClaw 用多 Agent 方式模拟一个完整 IT 团队，把 需求 → 开发 → 测试 → 交付，全程自动流转。

🤖 AI Workflow System - Turn your tasks into smart multi-agent workflows with specialized bots for different teams (Dev, PM, QA, etc.) that work together automatically

SWARM: System-Wide Assessment of Risk in Multi-agent systems

ContextAtlas — context infrastructure for AI coding agents: hybrid retrieval, project memory and retrieval observability via CLI, MCP server or embeddable library. Tree-sitter indexing, LanceDB vector search, FTS5 and token-aware context packing.

Local-first Rust daemon for AI work. Git-versioned memories, distilled wiki pages, sessions for Claude Code, Cursor, Codex, and any MCP client.

Keep your coding skills sharp in the age of AI

AI Agents summarizing YouTube videos - built with Neuron PHP AI framework.

Agent swarm simulation for structured world modeling with LLMs. Built on AgentOS.

🦞 Build on Crustocean

Automate the full pentest process with six Claude Code skills covering recon, exploit chains, and bug bounty report generation from any domain.