mcp
mcp
Fail
Health Warn
- License — License: MIT
- Description — Repository has a description
- Active repo — Last push 0 days ago
- Low visibility — Only 9 GitHub stars
Code Fail
- rm -rf — Recursive force deletion command in scripts/install-autogui.sh
Permissions Pass
- Permissions — No dangerous permissions requested
Purpose
This server provides a unified tool architecture for AI agents, offering capabilities like filesystem manipulation, shell execution, API requests, and code analysis. It is built using a TypeScript and Rust dual-runtime architecture to provide low-latency performance for system-level tasks.
Security Assessment
Overall Risk: High. This tool requires extreme caution because it is explicitly designed to access sensitive data and perform powerful system operations. It executes shell commands, directly reads and writes to the filesystem, makes outbound network requests, and can interact with cloud platforms. Additionally, a rule-based scan flagged a recursive force deletion command (`rm -rf`) inside an installation script. If an AI agent is manipulated or given unintended instructions, this tool could easily cause irreversible data loss or system compromise. No hardcoded secrets were found.
Quality Assessment
The project uses the permissive MIT license and shows active maintenance, with repository updates pushed as recently as today. However, it currently suffers from extremely low community visibility. With only 9 stars on GitHub, the project has not undergone the broad peer review or community testing typically expected for a tool that handles critical system operations.
Verdict
Use with caution.
This server provides a unified tool architecture for AI agents, offering capabilities like filesystem manipulation, shell execution, API requests, and code analysis. It is built using a TypeScript and Rust dual-runtime architecture to provide low-latency performance for system-level tasks.
Security Assessment
Overall Risk: High. This tool requires extreme caution because it is explicitly designed to access sensitive data and perform powerful system operations. It executes shell commands, directly reads and writes to the filesystem, makes outbound network requests, and can interact with cloud platforms. Additionally, a rule-based scan flagged a recursive force deletion command (`rm -rf`) inside an installation script. If an AI agent is manipulated or given unintended instructions, this tool could easily cause irreversible data loss or system compromise. No hardcoded secrets were found.
Quality Assessment
The project uses the permissive MIT license and shows active maintenance, with repository updates pushed as recently as today. However, it currently suffers from extremely low community visibility. With only 9 stars on GitHub, the project has not undergone the broad peer review or community testing typically expected for a tool that handles critical system operations.
Verdict
Use with caution.
Model Context Protocol server with 260+ tools for AI agents
README.md
@hanzo/mcp
Model Context Protocol (MCP) server with HIP-0300 unified tool architecture. TypeScript + Rust dual runtime.
Installation
npm install -g @hanzo/mcp
Quick Start
# Start with unified tools (default: 13 HIP-0300 tools)
hanzo-mcp serve
# Legacy individual tools
hanzo-mcp serve --legacy
# With UI extensions
hanzo-mcp serve --enable-ui
# List available tools
hanzo-mcp list-tools
# Install for Claude Desktop
hanzo-mcp install-desktop
HIP-0300 Tool Surface
13 canonical tools organized by axis. Each tool uses action-routed dispatch.
Core Tools (7)
| Tool | Axis | Key Actions |
|---|---|---|
fs |
Bytes + Paths | read, write, stat, list, mkdir, rm, mv, apply_patch, search_text |
exec |
Execution | run, background, ps, kill, logs |
code |
Symbols + Semantics | parse, search, transform, summarize |
git |
Diffs + History | status, diff, log, commit, branch, stash |
fetch |
HTTP/API | get, post, put, delete, download |
workspace |
Project Context | info, config, env, dependencies |
ui |
UI Components | list_components, fetch_component, search, install |
Optional Tools (6)
| Tool | Purpose |
|---|---|
think |
Structured reasoning |
memory |
Persistent storage |
hanzo |
Hanzo platform API (iam, kms, paas, commerce) |
plan |
Task planning |
tasks |
Task tracking |
mode |
Developer modes |
Usage
CLI Options
# Default: HIP-0300 unified surface (13 tools)
hanzo-mcp serve
# Legacy individual tools (read_file, write_file, bash, etc.)
hanzo-mcp serve --legacy
# Enable UI extensions
hanzo-mcp serve --enable-ui --enable-desktop
# Disable specific tools
hanzo-mcp serve --disable-tools plan,tasks
Programmatic Usage
import { getConfiguredTools } from '@hanzo/mcp';
// HIP-0300 unified surface (default)
const tools = getConfiguredTools({ unified: true });
// Legacy individual tools
const tools = getConfiguredTools({ enableLegacy: true });
// With UI extensions
const tools = getConfiguredTools({
unified: true,
enableUI: true,
enableDesktop: true,
});
AI Client Configuration
Claude Desktop / Cursor
Add to .mcp.json:
{
"mcpServers": {
"hanzo": {
"command": "npx",
"args": ["@hanzo/mcp", "serve"]
}
}
}
Architecture
src/tools/unified/ # HIP-0300 action-routed tools (fs, exec, code, fetch, workspace, hanzo)
src/tools/ # Individual tools (git, think, memory, tasks, plan, mode, etc.)
rust/src/tools/ # Rust native tools (exec, git, fetch, code, computer, etc.)
The Rust runtime provides native performance for latency-sensitive operations (<5ms clicks, <2ms keypress, <50ms screenshots).
Python SDK Parity
The Python implementation (hanzo-mcp on PyPI) exposes the same 13 HIP-0300 tools via entry-point discovery from hanzo-tools-* packages. Tool names and action schemas are identical across both runtimes.
pip install hanzo-mcp
License
MIT
Reviews (0)
Sign in to leave a review.
Leave a reviewNo results found