Custom agents

Self-hosted PaaS, like Heroku or Vercel on your own servers. Push your code, get a live app. Use one machine or many, even a laptop, all working as one cloud that stays yours.

cogito ergo sum: a creative workbench second brain (Obsidian + Claude Code) that organizes your work and improves with you

A desktop pet that knocks when your AI coding agent (Claude Code, Codex) is done or needs you. Windows, built with Tauri.

Design interfaces with your AI agent on a live canvas. Desktop app for Linux and Windows that works with Claude Code, Codex, OpenCode and Antigravity.

Turn a coding task into a workflow you can see: plan the steps, then watch Claude Code, Codex or Pi carry them out on a live diagram. Local-first desktop app.

Jesse Vincent's Superpowers skills (brainstorm, plan, TDD, debug, verify, review) adapted for GitHub Copilot agent mode, with project-scoped installers. MIT, adapted from obra/superpowers.

Context folding for DeepSeek Harness (DSH): wrap work in named tasks, fold finished spans into short summaries, recall originals on demand.

Watch and drive your Claude Code and Codex agents from your Mac and your phone. Native macOS app with worktrees, diff review and automations.

Plan the work. Let AI run it. An open-source, self-hosted project management platform for real businesses, installed from your browser, running on ordinary shared hosting. English and Arabic with full RTL.

Let an AI agent drive a jailbroken PS5 by itself: a virtual DualSense and fast screen capture that close the loop for unattended homebrew port development.

Your Multi-Agent Orchestrator for AI Governance in AI-assisted repos, and projects. 🏗️🧱

Design and review tools built for AI agents: 3D CAD, PCB, shaders, textures, and animation — code in, renders + machine-readable reports out.

Runs the base branch's tests against a PR's code and checks the description's claims against what actually ran. Signed evidence from CI.

Free subdomains under makes.fyi / agentdomains.co, as a Claude Code skill: agents claim a name and set DNS records, auto-HTTPS proxy or forwarding.

Claude Code mod: shows the critical decisions your agent made without being asked, so you can rethink them. Works in any kind of project.

An independent multi-model code reviewer that remembers the codebase between sessions

Local Git history intelligence for developers and coding agents — find precedents, explain code, trace regressions, and discover related tests.

A universal collection of reusable Loop Engineering kits for AI coding agents—not just prompts. Each kit combines Skills, Rules, Templates, workflows, and validation loops to help agents autonomously plan, execute, verify, fix, and improve engineering tasks.

Agent-first browser testing: deterministic replay, evidence packs, typed verdicts, and portable skills for coding agents.

Give 3D AI agents a voice and a face from any MCP client — text-to-speech, speech-to-text, audio-to-face lipsync (ARKit blendshapes), and the motion-capture clip library. Read-only over the live three.ws API; free NVIDIA lanes, no key required.

A collection of AI agent skills I have written

One-command Hermes setup for the Anycubic Kobra S1 with FreeCAD, OrcaSlicer, Moonraker and KlipperMCP.

Prospector IA Botz — funil de prospecção local (Google Places/Apify), redesign, deploy aaPanel e propostas.

Find files by what's in them: photos, PDFs, docs, voice memos and video, on your own computer. EmbeddingGemma 2 + llama.cpp; CLI, app and MCP server for agents.

Build and manipulate 3D scenes from natural language via AI agents.

MCP server for Zabbix 6.0-7.4 (monitoring): hosts, host groups, templates, items, triggers, problems and events, history and trends, maintenance, users, proxies, SLAs, script execution and configuration import/export. Read-only mode, tag-based tool filtering and rate limiting.

Open-source AI agent framework for automating the software development lifecycle (SDLC). Multi-agent orchestration for requirements, design, implementation, and QA.

Game sound effects and music as code: a synth DSL that exports WAV/OGG, looping chiptune songs and audio-sprite banks, with spectrogram previews and a live player. Sibling of vex.

AI-agent-first CLI for the Epic Games Store and Fab marketplace: search, claim free assets, manage your library, download.

Runtime permission firewall for AI agents - YAML policies, one-tap approvals (macOS popup/Telegram), tamper-evident signed audit log. Benchmarked: 103/103 attacks blocked, 8/8 normal dev ops allowed.

Codex skill for evidence-backed project handoffs across fresh tasks

The agentic workflow platform that builds itself. Describe what you need; the AI creates functions, agents and workflows, schedules them, and runs them durably on your own machine. Local-first Flutter desktop app with a Go sidecar.

Read-only performance diagnostics for DeepSeek Harness: session load/restore timing, spill-hit counts, compaction count and trigger, context-injection volume (AGENTS.md/skills/tool-schema token share), and LLM cache hit rate — surfaced via /fast, persisted as reconstructable session events with async sampling off the model path.

Persistent, structured project memory for Claude Code and Codex - event-sourced, typed, fail-open. Your agent stops re-deciding what you already decided.

Releases, screenshots and issue tracker for AI Devhouse: a desktop app that runs your Claude Code agents as an office. A VapuLab project.

Scrape public Telegram channels without API or login — Python CLI, MCP server for AI agents, dashboard

Manage your three.ws copy-trade follows from any AI agent — follow/unfollow leaders, tune sizing & guard rules, read the copy-intent inbox, and track fees owed. Account-scoped, non-custodial, over the live API.

A general-purpose reasoning and deep-research workflow for AI agents with semantic state, traceable evidence, validation, and change control

Open-source, self-hosted room for AI agent teams. Any model, Claude Code & Codex over ACP, its own computer in a container. Desktop + phone.

DOT pipeline for brainstorming, planning, TDD implementation, and multi-model code review on Attractor-compatible runners.

An AI agent flies ArduPilot over MAVLink. Every decision, command and fault streams to Voight as one trace.

Mirage - a browser extension that red-teams your AI like a real attacker. Autonomous recon→plan→test→judge→iterate agent, multi-provider, deterministic scoring. By KageX.

Todos and notes your AI agent can read and update. VS Code extension with a built-in MCP server for Claude Code, Cursor and any MCP client, plus GitHub Gist sync, a phone app, Markdown and Mermaid.

From Matt Pocock Skills to working code. Automate GitHub ticket development, verification, and repair with Codex.

Agent-readable knowledge base for reproducing and operating embodied-AI simulation toolchains. Five-layer docs (quickstart / concepts / field notes / troubleshooting / code), plus a cross-project layer of reproduction workflow, issue taxonomy, selection guide, best practices and terminology mapping.

Fast, minimalistic and very opinionated AI-first code editor

Find stale Git worktrees across all your repositories, see their disk usage and status, and clean up the ones you no longer need. Worktree cleanup for any coding agent or tool, including Claude Code, Codex and Cursor.

Browser-native 3D agent viewer and chat UI from three.ws: drop in a GLB, give it an LLM brain, and embed it anywhere.

Monocrawl MCP server and agent skill: live public data from 70+ platforms for AI agents

Not just a chatbot: a real coding agent on your Android phone. Inspect and edit files, run commands, and control apps and the UI over wireless ADB. No PC needed.

Raft Mobile: the Android and iOS client for Raft, built with Kotlin and Compose Multiplatform.

Simplified Technical English (ASD-STE100) skills for AI coding assistants, featuring extensible dictionaries and automated vocabulary discovery.

🧠 Statically compiled, ultra-light Go drop-in replacement for self-hosted Supermemory servers. Uses 1.5MB RAM, runs 100% offline with local Ollama, and executes semantic vector search in <30ms. Zero Node.js or WASM overhead.

Dynamic multi-agent workflow orchestration for the pi coding agent

MAX CLI для людей, ботов и агентов: личный аккаунт и Bot API MAX в терминале, MCP и навыки для Claude, Codex и Cursor

Adaptive extractive text compression for AI agents. Shorten academic documents without paraphrasing, with traceable edits, semantic review and optional lightweight OCR.

A cross-tool field guide to how I build with AI, with practical workflows and portable skills from research to maintenance.

Making multiple LLMs play Texas Holdem against each other

Agent Skills for real engineering work: impact maps, proof-driven development, production gates, investigations and more. For Claude Code and any Agent Skills-compatible agent.

Run short-lived Python, Node, and Bash with hard limits, live output, and verifiable receipts.

Stata CLI for AI agents and coding agents (Codex, Claude Code, Cursor, OpenCode). Run Stata 18/19/StataNow locally, execute .do files, inspect .dta data, and return structured JSON via a native Rust engine — no Python or PyStata required. 开箱即用的AI原生Stata命令行工具，附带完整skill，支持最新版stata。完全由Rust构建，无需Python依赖。

Read papers with Codex and Pi in Zotero's sidebar. Ask about passages and figures, revisit past conversations, and view interactive charts beside the PDF.

Agent control plane for governed AI coding: validate changes, enforce policy gates, track findings, proofs, and evals based on your habits.

Teach your engineering methodology once; every AI agent inherits it. Methods, gates, and durable memory, from one terminal board.

Agent Skills to help developers using AI agents with Lingui

A practical roadmap for engineering teams adopting agentic AI workflows and tools.

Portable offline AI assistant for SAP/ABAP learning, code drafting, debugging, and development planning. Runs locally with lightweight models, streaming responses, personal note retrieval, and optional GPU acceleration.

Production-grade Model Context Protocol (MCP) server for Bale Messenger with 32 tools for LLMs (Claude, Cursor, Antigravity) to interact with user accounts.

MCP server and CLI that let AI agents (Claude, Codex, any MCP client) see and drive a running WARUDO: scene, assets, character pose and IK, expressions, blueprints, camera and captures.

Create your own agentic analytics benchmark from your ClickHouse data-warehouse.

Open-source desktop AI companion with a VRM avatar, voice, long-term memory, computer use, and an evaluated general Agent Harness.

Sandbox for mini LLMs that run directly in the browser with 3D companion interactions and voice

🛒 MCP server + AI Telegram bot for Pyaterochka (5ka.ru): stores, products, prices & promos via Claude, ChatGPT or Telegram

GenPark skill: Utilize GenPark Workspace 4.0 for desktop automation, MS Office integration, and AI meeting notes.

GenPark AI Agent Skill - Multimodal ambient visual anchoring connecting physical AI Gadgets with digital Services inspired by Meta Ray-Ban / Muse.

Durable execution and guardrails for AI agents

Sandboxed macOS VMs with full Xcode for AI coding agents — plus a lightweight Linux container mode.

A Claude Code skill that turns a one-line idea into a published, evidence-backed page. Recon, real lab experiments, verification loops, two human checkpoints. Free and MIT.

Self-hosted web UI for Claude Code. Persistent, resumable chats on your own server — fork, rewind, and reach them from any browser.

Draws how your apps are built, every workflow and the data they store, from maps your coding agents write

Claude Code & Codex plugin for the open-source ArtCraft Craft App Suite: one craft Skill drives 13 local, headless Craft MCP apps (photo, vector, video, RAW, PDF, layout, sheets, docs, slides, CAD, audio, motion) with staged copies and verified exports.

Let your coding agent fix a prompt until it actually works, and prove it did. A self-hosted MCP server: cross-vendor judges, frozen baselines and memory.

Jev-powered MCP tool routing for Codex

Belay is an open-source, local-first security layer for AI coding agents (Claude Code, Codex, Cursor, OpenClaw, Hermes Agent and MCP) that blocks dangerous commands, secret leaks, and prompt injection at the tool-call boundary in under 100ms — no LLM in the decision path by default, no cloud, no phone-home.

AI assistant with a local Linux desktop, live browser view, and task replay.

🦆 [EMNLP 2026] Official Implementation of QUACK: Questioning, Understanding, and Auditing CollaborativeKnowledge in Multimodal Social Deduction Agents

Run the Hermes personal AI assistant inside a Cloudflare Sandbox container, fronted by a Cloudflare Worker. Single-tenant, BYOK.

Dralo Studio 短剧工坊 | 开源 AI 短剧创作工作台：剧本创作、无限画布、智能体、3D 分镜导演台、角色与场景管理、多模型图像/视频生成、多轨剪辑。本地运行，自备模型。Open-source AI short-drama studio with screenwriting, infinite canvas and storyboard workflows.

Trace a GX Works3 ladder without opening GX Works3 — read-only CLI + MCP server so anyone (or an AI agent) can find why a coil never turns on.

OMRON PLC AI integration, Program plc's with AI,

AI 驱动的 SIP 软电话客服 Agent —— Rust + PJSIP + Tauri + DeepSeek，支持实时 ASR/TTS 语音对话

Audit a Google Business Profile against 39 local SEO rules, get a client-ready HTML report, then fix it: rewrite the description, set holiday hours, reply to every review, post weekly with AI images, and get alerted when Google changes something. Python + Google Business Profile API + Claude. 207 offline tests.

A tiny friend in your Mac's notch that listens to your calls and whispers what to say. Live transcription on your Mac + your AI agent.

A grounded, local-first computer agent powered by Google Gemini with safe tools, bounded autonomy, and verifiable actions.

微信聊天记录误删恢复：从对方保留的单聊记录恢复自己的视角，找回与重要的人的对话。开源 Mac 工具，提供备份、恢复预览、完整 SOP 和 Agent 操作指南。实验版，需双方授权。WeChat chat history recovery for macOS.

🚀 Educational & minimalist DeepSeek Harness (DSH) architecture implementation powered by Cordis microkernel. 极简版 DeepSeek Harness 架构实现与学习指南。

DeepSeek Harness Web 鲸鱼娘桌宠插件：任务进展、会话费用、账户余额、峰谷调度与互动养成。

ShadeNet: anonymous Tor egress for AI agents, zk-proof gated. A local HTTPS proxy and MCP server for the agent; Shade Tree nodes provide the egress. Research preview on Sepolia.

CyberArena: Benchmarking AI Agents in Live Symmetric Attack-and-Defense Combat

在 Obsidian 中用本机 Agent 或模型 API 理解、改写笔记并审阅修改 | Chat with and edit your Obsidian vault using local agents or model APIs.