Mcp Defender — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited Mcp Defender (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
mcp-name: io.github.trickyfalcon/mcp-msdefenderkql
An MCP (Model Context Protocol) server for Microsoft Defender Advanced Hunting. Enables AI assistants to investigate security events using natural language by translating queries to KQL and executing them against Defender.
User: "Show me suspicious PowerShell activity in the last hour"
↓
AI translates to KQL using schema knowledge
↓
MCP executes query against Defender API
↓
AI interprets and explains the resultsAdvancedQuery.Read.All - Run advanced queriespip install mcp-msdefenderkql# Clone the repository
git clone https://github.com/trickyfalcon/mcp-defender.git
cd mcp-defender
# Create and activate virtual environment
python -m venv .venv
source .venv/bin/activate # On Windows: .venv\Scripts\activate
# Install dependencies
pip install -e ".[dev]".env.example to .envAZURE_TENANT_ID=your-tenant-id
AZURE_CLIENT_ID=your-client-id
# Option 1: Certificate authentication (recommended)
AZURE_CLIENT_CERTIFICATE_PATH=/path/to/combined.pem
# Option 2: Client secret authentication
# AZURE_CLIENT_SECRET=your-client-secretFor certificate authentication, combine your private key and certificate:
cat private.key cert.pem > combined.pemmcp-msdefenderkqlnpx @modelcontextprotocol/inspector mcp-msdefenderkqlAdd to your Claude Desktop config (~/Library/Application Support/Claude/claude_desktop_config.json):
{
"mcpServers": {
"defender": {
"command": "/path/to/mcp-defender/.venv/bin/python",
"args": ["-m", "mcp_defender.server"],
"env": {
"PYTHONPATH": "/path/to/mcp-defender/src",
"AZURE_TENANT_ID": "your-tenant-id",
"AZURE_CLIENT_ID": "your-client-id",
"AZURE_CLIENT_CERTIFICATE_PATH": "/path/to/combined.pem"
}
}
}
}| Tool | Description |
|---|---|
run_hunting_query | Execute KQL queries against Advanced Hunting |
get_hunting_schema | Get available tables and columns dynamically |
Once connected to Claude, you can ask:
// Find failed logon attempts
DeviceLogonEvents
| where ActionType == "LogonFailed"
| where Timestamp > ago(24h)
| summarize FailedAttempts = count() by AccountName, DeviceName
| top 10 by FailedAttempts
// Detect suspicious PowerShell
DeviceProcessEvents
| where FileName in~ ("powershell.exe", "pwsh.exe")
| where ProcessCommandLine has_any ("encodedcommand", "bypass", "hidden", "downloadstring")
| project Timestamp, DeviceName, AccountName, ProcessCommandLine
// Network connections to external IPs
DeviceNetworkEvents
| where RemoteIPType == "Public"
| where Timestamp > ago(1h)
| summarize ConnectionCount = count() by DeviceName, RemoteIP
| top 20 by ConnectionCount# Run tests
pytest
# Lint code
ruff check .
# Type check
mypy src
# Security scan
bandit -r srcThis server uses the WindowsDefenderATP API:
https://api.securitycenter.microsoft.comPOST /api/advancedqueries/runMIT
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.