About
Model Context Protocol (MCP) server collection and configurations
README
Personal collection of tools for AI coding agents — MCP servers, CLIs, skills, and wrappers.
Each tool lives as a git submodule under this repository. Some are original projects (AES-style architecture); others are local forks or wrappers around well-known upstream projects. Most expose an MCP interface, but the collection is not limited to MCP — anything that makes agents more capable belongs here.
Tools
| Submodule | Purpose | Stack |
|---|---|---|
| blender-arwaky | Control Blender 3D (scenes, objects, render, Python) | Python |
| cloudmail-arwaky | Virtual email / inbox on Cloudflare Workers | TypeScript |
| contex7-arwaky | Up-to-date library docs (Context7) | TypeScript |
| fetch-arwaky | Fetch web content (HTML, Markdown, JSON, YouTube transcripts) | TypeScript |
| lean-arwaky | LeanCTX — context engineering / token compression for agents | Rust |
| lint-arwaky | Architecture linter (AES rules) for Rust, Python, TypeScript | Rust |
| testing-arwaky | Autonomous test engine with self-healing | Python |
| vision-arwaky | Computer vision — image, OCR, video, tracking, visual memory | Python |
| ponytail-arwaky | Lazy senior dev mode — AI agent wrapper (Ponytail) | TypeScript / Node |
| codegraph-arwaky | Semantic code intelligence — Rust kernel, local index, MCP | Rust / TypeScript |
| qwen-web-arwaky | Qwen AI Web Automation CLI & MCP Server | Python / Playwright |
| caveman-arwaky | Caveman — token compression skill + MCP middleware | TypeScript |
| graphyarwaky | Graphify wrapper — query knowledge graphs via MCP | TypeScript / Node |
| anytype-arwaky | Anytype MCP wrapper — manage objects/lists via AI | TypeScript / Node |
What each tool does
blender-arwaky
Bridges Blender 3D to MCP clients. Control scenes, import assets, render, and run Blender Python through a small set of universal tools. Requires Blender 3.0+ and a Blender addon.
cloudmail-arwaky
Email management on Cloudflare Workers: virtual users, inbox, settings, API keys. Surfaces: HTTP API, CLI, MCP (Hydra meta-tools), and Web UI.
contex7-arwaky
Local wrapper around Context7. Resolves library IDs and pulls version-specific documentation and code examples into the agent context (avoids hallucinated APIs).
fetch-arwaky
Local wrapper around mcp-fetch-server. Fetch URLs as HTML, Markdown, plain text, JSON, readable article content, or YouTube transcripts. Includes SSRF protection and response size limits.
lean-arwaky
LeanCTX — context engineering layer for coding agents. Decides what agents read, compresses prompts (often 60–90% fewer tokens), remembers session knowledge, and guards filesystem access. Local-first Rust binary.
lint-arwaky
Architecture linter for Rust, Python, and TypeScript. Enforces 24 AES rules (naming, imports, quality, layer roles, orphans), bridges external linters (Clippy, Ruff, ESLint, …), and exposes a 5-tool MCP server with full CLI parity.
testing-arwaky
Autonomous test engine for AI agents: run tests with self-healing, AST analysis, coverage audit, synthetic data generation, and pytest workflows.
vision-arwaky
Unified computer-vision MCP: image analysis (VLM), OCR (Tesseract), video (ffmpeg/OpenCV), object tracking, and perceptual-hash visual memory.
ponytail-arwaky
Wrapper repository that integrates the Ponytail project — "lazy senior dev" AI agent plugin. Provides MCP server built with CLI scripts and optional ready-made MCP configs (ponytail_arwaky_local.json, ponytail_arwaky_remote.json).
qwen-web-arwaky
Qwen AI Web Automation CLI & MCP Server. Automates browser interactions with chat.qwen.ai using Playwright, supporting batch prompt processing, real-time file watching, persistent session management, and 1:1 MCP Server tool integration for local AI agents.
caveman-arwaky
Caveman — token compression skill for AI coding agents. Makes agents talk like caveman: 65% fewer output tokens, full technical accuracy. Includes caveman-shrink MCP middleware that wraps any MCP server and compresses its tool descriptions. Supports Claude Code, Codex, Gemini, Cursor, Windsurf, Cline, Copilot, 40+ agents.
graphyarwaky
Wrapper around Graphify. Turns codebases and docs into queryable knowledge graphs and exposes query, path, and explain through MCP.
Repository layout
agents-arwaky/
├── blender-arwaky/ # original
├── cloudmail-arwaky/ # original
├── contex7-arwaky/ # Context7 wrapper
│ └── context7/
├── fetch-arwaky/ # fetch-mcp wrapper
│ └── fetch-mcp/
├── lean-arwaky/ # LeanCTX wrapper
│ └── lean-ctx/
├── lint-arwaky/ # original
├── testing-arwaky/ # original
├── vision-arwaky/ # original
├── ponytail-arwaky/ # Ponytail wrapper
│ └── ponytail/
├── qwen-web-arwaky/ # original
├── caveman-arwaky/ # Caveman wrapper (token compression)
│ └── caveman/ # upstream (JuliusBrussee/caveman)
├── graphyarwaky/ # Graphify wrapper
│ └── graphify/ # upstream Graphify tooling
├── .gitmodules
└── README.md # this file
Wrapper submodules typically contain:
- Upstream (or vendored) source under a nested directory
script/install.sh— local build intodist/- Optional
*_local.json/*_remote.json— MCP client config snippets
Clone and init
git clone --recurse-submodules https://github.com/rakaarwaky/agents-arwaky.git
cd agents-arwaky
If you already cloned without submodules:
git submodule update --init --recursive
Update all submodules to their tracked commits:
git submodule update --remote --merge
Agent / MCP client configuration
Point your agent host or MCP client (Qwen Code, Claude Desktop, Claude Code, Cursor, etc.) at each tool. Patterns differ by stack:
Python (uv)
{
"mcpServers": {
"blender-arwaky": {
"command": "uv",
"args": ["--directory", "/path/to/agents-arwaky/blender-arwaky", "run", "blender-mcp"]
}
}
}
Rust binary
{
"mcpServers": {
"lint-arwaky": {
"command": "lint-arwaky-mcp"
}
}
}
Go binary (local build)
{
"mcpServers": {
"github": {
"command": "/path/to/agents-arwaky/github-arwaky/dist/github-mcp-server",
"args": ["stdio"],
"env": {
"GITHUB_PERSONAL_ACCESS_TOKEN": "<token>"
}
}
}
}
Node / npx (wrappers)
{
"mcpServers": {
"fetch": {
"command": "node",
"args": ["/path/to/agents-arwaky/fetch-arwaky/fetch-mcp/dist/index.js"]
}
}
}
Caveman-shrink MCP middleware
{
"mcpServers": {
"caveman-shrink": {
"command": "node",
"args": ["/path/to/agents-arwaky/caveman-arwaky/dist/caveman-shrink/index.js"]
}
}
}
Several wrappers ship ready-made snippets (*_local.json, *_remote.json) — prefer those when present.
License
Each submodule has its own license (typically MIT or Apache-2.0). See the LICENSE file inside each project.
Installing Arwaky
This server has no published package — it is built from source. Open the repository and follow its README.
▸ github.com/rakaarwaky/mcp-arwakyFAQ
Is Arwaky MCP free?
Yes, Arwaky MCP is free — one-click install via Unyly at no cost.
Does Arwaky need an API key?
No, Arwaky runs without API keys or environment variables.
Is Arwaky hosted or self-hosted?
Self-hosted: the server runs locally on your machine via the install command above.
How do I install Arwaky in Claude Desktop, Claude Code or Cursor?
Open Arwaky on unyly.org, pick your client tab (Claude Desktop, Claude Code, Cursor) and press Install — the config is generated automatically, no JSON editing.
Related MCPs
GitHub
PRs, issues, code search, CI status
by GitHubFilesystem
Secure file operations with configurable access controls.
Memory
Knowledge graph-based persistent memory system.
Template MCP Server
A CLI tool to create a new Model Context Protocol server project with TypeScript support, dual transport options, and an extensible structure
by mcpdotdirectAmap Maps Mcp Server
MCP server for using the AMap Maps API
by duxiaohuiSupabase
Database, auth and storage
by SupabaseEverything
Reference / test server with prompts, resources, and tools.
Git
Tools to read, search, and manipulate Git repositories.
Sequential Thinking
Dynamic and reflective problem-solving through thought sequences.
Time
Time and timezone conversion capabilities.
Compare Arwaky with
Not sure what to pick?
Find your stack in 60 seconds
Author?
Embed badge for your README
Browse similar
All development MCPs
