Custom agents

DeepSeek Harness（DSH）第三方插件精选目录：一键安装，条目均指向插件作者原仓库。

Claude Code | Codex plugin for React Native — explores codebases, designs architecture, implements features and QA the code in provable way

Native Windows GUI for Oh My Pi: sessions, models, providers, MCP, plugins, context, and browser tools.

Private AI WYSIWYG Markdown editor with agentic chat & voice dictation. Open source, no cloud, no tracking.

A React + TypeScript spatial canvas and node-board library for whiteboards, visual editors, and graph interfaces.

Complete 21-phase local business SEO audit plugin for Claude Code — 24 skills, 36 commands, 6 sub-agents, AI visibility, competitor analysis, and actionable roadmaps

Local agent skill for ROS 2 robot control

Blue: a TUI is not a package, it is a Cordis plugin tree — a modern terminal UI for DeepSeek Harness with hot-swappable render, interaction, and command plugins.

A Retrieval-Augmented Generation (RAG) application for querying legal documents. It uses PostgreSQL, Elasticsearch, and LLM to provide summaries and suggestions based on user queries. Features data ingestion with Airflow, real-time monitoring with Grafana, and a Streamlit interface.

Wenshu（文枢）— 面向人文社科研究的 AI 知识处理工作流：本地知识库 / RAG / 中文引文 / 理论谱系。AI knowledge workflow for humanities & social sciences: local knowledge base, RAG, citation (GB/T 7714), knowledge graph.

A complete, verified guide to becoming an AI Product Manager and passing AI PM interviews: LLM and ML foundations, evals, RAG, agents, product craft, 7 industry domains, sourced case studies, a 160+ question interview bank, and career resources. Every fact sourced, no hallucinations.

DSH-native terminal workbench (TUI) for DeepSeek Harness with durable sessions, structured tools, plans, subagents, and rotating Pokémon startup art.

The pit crew for agentic coding.

A curated collection of resources for building “AI Scientist” systems: AI that assists scientific discovery through literature intelligence, hypothesis generation, experiment planning, tool-use, evaluation, and scientific communication. 🌟 Star if you like it!

A shared kanban board for people and AI agents in DeepSeek Harness.

🍳 Vibe Cook — 跟着做就会。把 HowToCook 做成沉浸式一步步烹饪 App。

Agent Runtime Platform — the open-source runtime that powers AI employees

Portal - public MCP endpoint

A typed, replayable, budget-aware programming language for LLM agents. Compiles to Python & TypeScript.

Humanlike — deterministic persona, context, memory and privacy controls for conversational AI agents · Детерминированное поведение ИИ-агентов

J.A.D.E. — a text-based chatbot built on the Google Gemini API, with a Python backend and an HTML/CSS interface.

Spec Kit + Roo Code integration starter: one-command setup, 7/10/17-mode AI teams. Roo Code sunset May 2026 - Zoo Code (community successor) compatibility untested.

A service-agnostic proof-of-intelligence gate. Issues escalating challenges and mints offline-verifiable proofs to tell intelligent agents from dumb scripts

A long-horizon agent orchestration toolkit in OCaml implementing Cordis-like extensibility and graceful fallbacks

AI Terminal Manager — unified session history for Claude Code, Codex, Gemini CLI, opencode and Pi, plus a persistent tmux sidebar

A local-first job-search assistant: watches your email, chats and calendar, finds roles worth your time, and surfaces the few things that need you today. Never applies or sends without approval.

AI agents should not make irreversible changes. TraceFold escrows the inverse before an effect lands, or refuses it, and issues receipts anyone can verify offline.

使用 Google Gemini 总结 YouTube 视频。向 Gemini 询问视频内容，享受双语同步字幕，并导出转录文本。具有原生 YouTube 风格的界面。请从 greasyfork 安装最新版本。 Summarize YouTube videos with Google Gemini. Ask Gemini about video content, enjoy bilingual synced subtitles, and export transcripts. Features a native YouTube-style UI.Please install the latest version from the greasyfork.

Managing prompts，, skills, MCP tools & wallpaper for DeepSeek Harness

A multilingual Agent Skill Pack that turns AI agents into guided learning coaches.

Hive mind mechanics for agents. A step closer towards AGI

Your AI forgets everything between sessions. This fixes that — 98%+ retrieval accuracy, 100% on LongMemEval, 99% token savings. 44 MCP tools. Fully local, zero cost.

Agent-native task management for humans and AI agents, built around a local-first API, SQLite storage, projects, calendar views, backups, and plugin-ready integrations.

Multi-language static analysis for agents, editors, and large repositories.

A source-aware directory of Embodied AI agents, LLM/VLM systems, VLA models, and robotic manipulation projects.

ArtiControlNet——辅助设计的 AIGC 系统

Clone-and-go agent skills + a starter Obsidian vault for an AI-maintained personal wiki. Based on Karpathy's LLM Wiki pattern.

😼 AI-участница Telegram-чата с памятью, локальными мемами, реакциями и собственным характером

Evidence-backed audits, evaluations, and measured improvements for any AI agent.

Research design for a Jev-native agent system: tool integration, speculative parameter proposals, external helper logits Top-k proposals with Jev-controlled fallback ,decision-aware hierarchical memory, and dependency-aware replanning.Feature:Jev naturallanguage conversation prototype using external helper logits and dynamic Top-k token selection.

Rwally is the AI agent trading index on Arc. Members pool USDC into spot crypto index baskets and ratify every rebalance by on-chain vote. Immutable contracts: no proxy, no pause, no upgrade, no admin key.

A structured, continuously updated atlas of post-training research across language, agents, multimodal models, generative media, and embodied intelligence.

Standalone MCP server for Obsidian vaults — hybrid search, structured memory, task management, files, OAuth 2.1. Run locally, self-host, or one-click deploy for remote access.

Voice agents in YAML, compiled to Pipecat or LiveKit. Production code you own, deployed anywhere.

Zero-trust security gateway between AI agents and MCP servers: scoped capabilities, redaction, rate and cost limits, and verifiable session logs.

Guided Pi skill for a secure persistent mobile AI-agent environment on macOS and Windows with WSL2.

Test mode for your integrations, built for the way agents build

Python video editing with JSON plans, local AI, and MCP tools.

Free web search and page fetching for agents. Hosted MCP, no API key required.

Discovery Foundation Models (DFMs): From solving tasks to discovering new knowledge.

A Jev-powered resolver for agent harnesses to find deterministic commands and their context in a nested capability tree.

Trace how a silent tool failure taints a whole agent run: dependency graph, taint paths, and grounded or tainted verdicts.

pyTPG — A modern, modular Python framework for Tangled Program Graphs. Install with `pip install pytpg`.

Reusable AI skills for Codex, Claude Code & Copilot CLI — Markdown-only, safety-first samples for diagnosing bugs, drafting changelogs, and pre-merge checks.

Agent CLI Framework: self-describing command-line tools for AI agents, built on oclif

Consensus-driven multi-agent emergency intelligence system using computer vision, real MCP protocol, and iterative AI reasoning for trusted operational decision-making.

Windows 微信对话分析助手：本地读取、脱敏、TypeSafe Jev 判断与加密历史

Evidence-backed attack paths, drift, and cost auditing for Hetzner Cloud.

A terminal where one Claude Code session supervises the AI agents in your other tabs: which is stuck, which is done, which is still running. Group chat, task panel and MCP tools for multi-agent work. Built on Ghostty.

Daedalus CRM — 跑在你自己电脑上的 AI CRM。Mac / Windows 桌面端：个人版免费，团队版按席位收费、客户数据加密同步。AI-native CRM desktop app for macOS & Windows — free for individuals, per-seat team plan with encrypted sync. Open source (AGPL-3.0).

A curated list of projects, integrations, and resources for Jev, TypeSafe AI's System One model.

Local-first MCP server and CLI in Rust: any page as LLM-ready Markdown, whole-site crawls, keyless search, and local captcha solving. No cloud, no API keys.

Fine-tune Qwen2.5-Coder-7B for free (QLoRA on Kaggle T4) and run a local coding agent that runs your code instead of guessing — 8 GB Mac, llama.cpp, LoRA weight analysis with charts

Lab Agent — open-source coding agent (terminal + desktop)

A sandbox-first JavaScript engine for AI agents: host-set capabilities and limits, deterministic runs, TypeScript annotations, test262 on every pull request.

Reflect helps engineering teams see whether AI coding agents are actually improving delivery or just burning tokens, failing tools, and creating hidden debt

Open-source agent reliability platform for discovering, investigating, and fixing production agent failures

Your AI agent, inside ETABS: an MCP server to model, load, analyse and design structures through the official ETABS API.

Git-versioned, contiguity-checked memory for AI agents: verbatim transcripts, byte-offset recall, no LLM

Chessmata: chess game for humans and AI agents

Phone Agent: iOS AI assistant with ESP32-S3 BLE mouse, keyboard and GPIO48 RGB firmware

Open-source, self-hosted AI employees that ask before they act. 46 business roles, approval-gated autonomy, untrusted-content safeguards, sandboxed execution, Graph RAG and token budgets. Expanding across memory, voice, security and physical AI. WhatsApp, Telegram, Discord and 20 LLM providers. MIT licensed.

Live Aeon instance: the growth agent for MiroShark and $MIROSHARK, running skills on GitHub Actions cron.

A voice remote for AI agents: push-to-talk, procedural avatar + live status ring on the M5Stack StopWatch (ESP32-S3 round AMOLED). Hermes gadget firmware, hub plugin, desktop sim, WiFi OTA.

Indian Stock Research MCP server — turn Claude into a personal equity analyst powered by live Screener.in data

Asking for What Was Never Requested: horizontal and vertical proactivity in LLM agents. Need-graph metrics (no LLM judge) and Q&D, which trains a questioner from the consequences of its questions.

Carefully human hand written llm jailbreaks for the latest models that are available on the market. GPT-6 Astra, GPT-6.1 SoL, Opus 4.7, Opus 4.6, Sonnet 4.6, Haiku 4.5, Deepseek V4, Deepseek Flash, DeepSeek V4.1 Pro / Flash, Qwen 3.8 MAX, Qwen 3.8 Omni-Flash, Qwen 3.7 etc

Open-source, local-first AI team workspace. Several AI models work together in one conversation, each with its own provider, model and tool whitelist. Android and VS Code clients pair in. No telemetry.

Amalia Android Assistant

Unofficial Python wrapper for the WhatsApp Cloud API (Meta): send text, media, locations, contacts, lists, templates and reactions, and parse webhooks

Crossroad A2A: agent-to-agent mail so ChatGPT, Claude, Gemini, Codex and Grok agents work as one team across machines and SaaS apps. Open source, v0.1 beta.

Swarm intelligence for distributed AI agents. One agent learns, the whole swarm knows. Shared memories, reminders and agent-to-agent messages, ranked by trust. MCP and Claude Code ready.

A fast macOS terminal for working with coding agents, built on libghostty.

一体化AI智能订阅管理面板 —— 订阅管理 · 组合拼接 · 多格式转换（Clash / sing-box / Shadowrocket…）· 规则与文档托管 · AI 助手。Cloudflare 一键部署，也可 Docker 自建，数据完全自持。AGPL-3.0

Ashley 桌面语音助手 · 类贾维斯（Jarvis）AI 助手 · 全双工实时语音 + 3D 头盔 · macOS 开源 | macOS desktop voice assistant with a 3D avatar

How to use pstack without Cursor: universal workflow for Claude Code (pstack-claude), Pi (pi-pstack), Codex, & Gemini. Zero dependencies, v0.15.0+.

A locally powered RAG platform for exploring and answering questions about Technical University of Kenya information using retrieval, semantic search, and LLMs.

Windows utility focused on llm performance profiles, generation speed, and low vram profile for local AI workflows.

Windows utility focused on vram optimization, llm memory usage, and gpu memory budget for local AI workflows.

Multi-tenant RAG reference architecture (Next.js, Supabase, Pinecone): per-tenant isolation, hybrid search, cited answers, LLM failover, and a public hallucination benchmark.

A brain for your mail: MailFathom turns IMAP mailboxes into a self-hosted, AI-native service. Mail synchronizes into your own PostgreSQL, is indexed for search and retrieval, and is served to AI agents over the Model Context Protocol. Semantic retrieval, answering, and gated write tools. .NET 10, AGPL-3.0-only.

原神千星奇域 MCP 服务器：让 Claude、Cursor 等 AI 替你搭关卡、写节点图、自动试玩 | Unofficial MCP server for Genshin Impact's Miliastra Wonderland (UGC) editor: AI builds scenes, node graphs and playtests levels

The open-source agentic editor, built on the VS Code core.

AI-powered Telemed Chatbot utilizing GraphRAG architectures for enhanced medical knowledge mapping and hallucination-free retrieval.

Background task queue for Claude Code and other AI CLIs — web UI + Telegram bot

Find what's using a port, who started it (Claude Code, Cursor, Docker…), whether your network can reach it, and stop it safely. An lsof alternative for developers: TUI, CLI and a live harbour view.

AI workspace for shipping GitHub changes and managing VPS infrastructure with prompt-to-edit flows, code review, project summaries, PR actions, and live SSH monitoring tools.

A QA engineer for your coding agent. It tests each React Native change on a simulator and tells you PASS or FAIL, with video proof.

An evidence-first harness for agent-assisted research

Curated open-source small tools and workflow recipes for Chinese e-commerce/retail: Douyin/Xiaohongshu/WeChat Channels scraping, Taobao/1688/PDD/JD ops, WeChat SCRM, live-streaming, logistics, Chinese NLP, cross-border. Universal tools kept and tagged.