Atlas / MCP servers / zhangzhongnan928 / Personal Assistant Agent

Personal Assistant AgentSAFE

Name: Personal Assistant Agent
Author: zhangzhongnan928

mcp/zhangzhongnan928/personal-assistant-agent

MCP server by zhangzhongnan928 · Audited 2026-10-09

⚑ Report / re-scan Source ↗

A personal assistant AI agent built with the Model Context Protocol (MCP)

Verdict

SAFE

Grade

Trust score

86 /100

Exposed tools

19 12r · 6w · 1d

Transport

stdio

License

—

Stars

Overview

From the repository's own README, as read at the audited commit. Badges and raw HTML are left out.

A versatile personal assistant AI agent built with the Model Context Protocol (MCP) that helps with calendar, tasks, emails, and more.

Overview

This project is a Model Context Protocol (MCP) server that provides a set of tools for a personal assistant agent. It can be integrated with MCP clients like Claude for Desktop to give AI assistants the ability to:

Manage calendar events
Track tasks and to-dos
Read and send emails
Search the web and retrieve information
Control smart home devices

Requirements

⚠️ IMPORTANT: Python 3.10 or higher is required for the MCP SDK. The server will not work with earlier Python versions.

Python 3.10+
MCP SDK 1.2.0+
Required Python packages (see requirements.txt)

Installation

Clone the repository:

git clone https://github.com/yourusername/mcp-pa-ai-agent.git
cd mcp-pa-ai-agent

Ensure you have Python 3.10+:

python --version

If your system Python is older than 3.10, set up a compatible environment:

# Using conda
conda create -n mcp-env python=3.10
conda activate mcp-env

# OR using venv (if Python 3.10+ is installed elsewhere)
python3.10 -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

Install dependencies:

pip install -r requirements.txt

Configure environment variables by copying the example file:

cp .env.example .env

Edit the .env file with your API credentials and settings.

Running the Server

Start the MCP server with:

python mcp_server.py

The server will start and listen for MCP client connections.

Connecting to Claude for Desktop

Install Claude for Desktop

Configure Claude for Desktop to use this MCP server by editing the configuration file at:
MacOS/Linux: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%\Claude\claude_desktop_config.json

Add the

Read from source at commit fec8e044b3cfOBSERVED · 2026-10-09

Connect

Built from this server's own package name, version and transport as found in its source — not copied from anyone's documentation, so it cannot drift against a page we do not control. Replace the environment placeholders with a token scoped to the least it needs.

claude-code

claude mcp add mcp-pa-ai-agent --env DUCKDUCKGO_API_KEY=${DUCKDUCKGO_API_KEY} --env GOOGLE_ACCESS_TOKEN=${GOOGLE_ACCESS_TOKEN} --env GOOGLE_CLIENT_SECRET=${GOOGLE_CLIENT_SECRET} --env GOOGLE_REFRESH_TOKEN=${GOOGLE_REFRESH_TOKEN} -- npx -y [email protected]

claude-desktop

{
  "mcpServers": {
    "mcp-pa-ai-agent": {
      "command": "npx",
      "args": [
        "-y",
        "[email protected]"
      ],
      "env": {
        "DUCKDUCKGO_API_KEY": "${DUCKDUCKGO_API_KEY}",
        "GOOGLE_ACCESS_TOKEN": "${GOOGLE_ACCESS_TOKEN}",
        "GOOGLE_CLIENT_SECRET": "${GOOGLE_CLIENT_SECRET}",
        "GOOGLE_REFRESH_TOKEN": "${GOOGLE_REFRESH_TOKEN}"
      }
    }
  }
}

Exposed tools (19)

12 read · 6 write · 1 destructive. Blast radius: 1 tool can delete or overwrite — an agent that can be talked into calling a tool can be talked into calling this one.

Tool	Risk	Description
`add_task`	write	Add a new task.
`control_device`	read	Control a smart home device.
`create_chart`	write	Create a chart from data points.
`create_event`	write	Create a new calendar event.
`create_scene`	write	Create a new scene with defined device states.
`delete_task`	destructive	Delete a task.
`get_device_state`	read	Get the current state of a smart home device.
`get_emails`	read	Get recent emails from a specific label.
`get_events`	read	Get upcoming calendar events.
`get_free_time`	read	Find free time slots in your calendar for a specific date.
`get_news`	read	Get latest news, optionally filtered by topic.
`get_weather`	read	Get current weather information for a location.
`list_devices`	read	List all available smart home devices.
`list_tasks`	read	List tasks with optional filtering by status.
`read_email`	read	Read the full content of a specific email.
`search_emails`	read	Search emails using Gmail search syntax.
`send_email`	write	Send an email.
`update_task_status`	write	Update the status of a task.
`web_search`	read	Search the web for information.

Trust audit

SAFEgrade B · trust 86/100 Nothing in the source contradicts what it says it does. Grade A is reserved for packages that have also passed the behavioural sandbox.

Layer	What it checks	Result
L0	Provenance & inventory	PASS
L1	Static analysis of the code	PASS
L2	Instruction surface (what it tells the agent)	PASS
L3	Class-specific surface	WARN
L4	Behavioural (sandbox)	SKIPPED

What the source does

Filesystem: declared (6 observation(s))
Network: declared (5 observation(s))
Shell: none-observed
Dependencies: not all pinned
Secrets in source: none-found

Findings (12)

MEDIUMFilesystem / path · mcp.destructive_tools · CWE-22, CWE-59

delete_task

Why it matters. 1 tool(s) can delete or overwrite

Fix. prefer a read-only mode or scoped tokens; the page states the blast radius

LOWInventory / provenance · inv.no_license · CWE-1104

Why it matters. no LICENSE file and no repo licence

Fix. add a licence