MinesweeperSAFE
An MCP server for playing Minesweeper
Overview
From the repository's own README, as read at the audited commit. Badges and raw HTML are left out.
This is an Model Context Protocol server that allows an MCP client agents to play a game of Minesweeper. It is intended to be run alongside the Minesweeper game server.
View the entire video demo at https://youtu.be/CXXMafVtlEQ (16x speedup).
Getting started
- Follow the instructions of the game server to start it locally.
- Build the MCP server:
npm install npm run build
- Configure your MCP client to add the tool. For example, here is how to add the tool to Claude Desktop on Windows's
claude_desktop_config.json(locating the file), assuming you cloned the repo atC:\path\to\repo\minesweeper-mcp-server:
{
"mcpServers": {
"mcp-server": {
"command": "node",
"args": ["C:\\path\\to\\repo\\minesweeper-mcp-server\\build\\index.js"],
"env": {
"DEBUG": "*"
}
}
}
}
- Claude Desktop : Restart Claude Desktop to let it pick up the tools. Be sure to quit from the tray menu icon, not from the app (which simply hides the window). If you click the Tools icon, it should show the new tools:
Example prompt
Start a new game of Minesweeper. Try your best to keep playing until you have flagged all mines. Remember that the coordinates are 0-indexed.
Example interaction
The actual conversation is very long. Here are some snippets:
Game start
Placing flag at the wrong place

2 read · 1 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 |
|---|---|---|
click | read | Click at a cell on the Minesweeper board |
flag | read | Place a flag at a cell on the Minesweeper board |
start_game | write | Start a new game of Minesweeper |
unflag | destructive | Remove the flag at a cell on the Minesweeper board |
Trust audit
SAFEgrade B · trust 89/100 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 (1 observation(s))
- Network
- declared (2 observation(s))
- Shell
- none-observed
- Dependencies
- not all pinned
- Secrets in source
- none-found
Findings (5)
unflag
@types/node, typescript
static/screen_recording.gif
static/screen_recording.mp4
Gates applied: no_behavioural_pass, no_license.
f0179de43e37full audit observations/trust-audit/mcp-server/tonypan2__minesweeper.json · Report an issue / request a re-scanAudit history
Every audit this server has had. A grade with a past is a grade somebody is still checking.
| Date | Source | Verdict | Grade | Score | Change |
|---|---|---|---|---|---|
| 2026-10-07 | f0179de43e37 | SAFE | B | 89 | first audit |
Questions
What is the Minesweeper MCP server?
An MCP server for playing Minesweeper
What tools does Minesweeper expose?
4 in total: 2 read-only, 1 that write, and 1 that can delete or overwrite (unflag). Every one is listed on this page with its risk.
Is Minesweeper safe to connect to an agent?
The audit found nothing in the source that contradicts what it says it does, and graded it B (89/100). Grade A is held back for packages that have also passed a sandboxed behavioural run, which is why a clean server reads B. Separately from the audit: 1 of its tools can destroy data, so scope the token you give it to what you actually need.
What credentials does Minesweeper need?
No credential environment variables were found in its source, so it appears to need none.
How does Minesweeper run?
It speaks stdio, so it runs as a local process your client starts. It is published on npm as minesweeper-mcp-server at 0.1.0.
How current is this page?
The grade is for one exact copy of the source (f0179de43e37), read on 2026-10-07. The repository is watched and re-audited when it changes.