Security Scanner MCP Server
Integrates with Nuclei vulnerability scanner from ProjectDiscovery's GitHub repository and references other GitHub-based MCP implementations
Provides Markdown as an output format option for security scan reports
Uses Pydantic for data validation and settings management in scan configurations and results
Implements security scanning functionality using Python 3.10+ with asyncio for concurrent vulnerability assessments
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@Security Scanner MCP Server scan example.com for critical vulnerabilities"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Security Scanner MCP Server
Comprehensive security scanning using Nuclei vulnerability scanner with cluster-wide capabilities.
Part of the Agentic System - a 24/7 autonomous AI framework with persistent memory.
Comprehensive security scanning integration using Nuclei vulnerability scanner with cluster-wide capabilities.
Features
Single Target Scanning: Scan individual URLs or IPs with customizable severity and templates
Network Scanning: Batch scan multiple targets or CIDR ranges
Cluster Integration: Scan all nodes in the agentic cluster
Template Management: List, filter, and update Nuclei templates
Scan History: Retrieve and analyze previous scan results
Scheduled Scans: Configure periodic security assessments
Multiple Output Formats: JSON, JSONL, or Markdown reports
Related MCP server: Guardian Ops
Installation
cd ${AGENTIC_SYSTEM_PATH:-/opt/agentic}/mcp-servers/security-scanner-mcp
source ${AGENTIC_SYSTEM_PATH:-/opt/agentic}/.venv/bin/activate
pip install -e .MCP Configuration
Add to ~/.claude.json:
{
"mcpServers": {
"security-scanner": {
"command": "${AGENTIC_SYSTEM_PATH:-/opt/agentic}/.venv/bin/python",
"args": ["-m", "security_scanner.server"],
"env": {
"NUCLEI_BIN": "${HOME}/go/bin/nuclei",
"SCAN_RESULTS_DIR": "${AGENTIC_SYSTEM_PATH:-/opt/agentic}/security-scans"
}
}
}
}Available Tools
scan_target
Scan a single target with Nuclei vulnerability scanner.
Parameters:
target(required): URL or IP addressseverity: Array of severity levels ["info", "low", "medium", "high", "critical"]templates: Specific template paths or tagsrate_limit: Requests per second (default: 150)timeout: Scan timeout in seconds (default: 300)output_format: "json", "jsonl", or "markdown"
Example:
{
"target": "https://example.com",
"severity": ["high", "critical"],
"rate_limit": 100
}scan_network
Scan multiple targets from a list or CIDR range.
Parameters:
targets: Array of target URLs/IPstarget_file: Path to file with targets (one per line)severity: Severity filterparallel: Number of concurrent scans (default: 5)rate_limit: Requests per second per target
scan_cluster_nodes
Scan all nodes in the agentic cluster for vulnerabilities.
Parameters:
scan_type: "network", "web", "api", or "full"severity: Severity levels to check
list_templates
List available Nuclei templates by tag, severity, or author.
Parameters:
tag: Filter by tag (e.g., "cve", "exposure")severity: Filter by severity levelauthor: Filter by template author
update_templates
Update Nuclei templates to the latest version.
get_scan_results
Retrieve results from a previous scan by scan_id.
Parameters:
scan_id(required): Scan identifierlimit: Max results to return (default: 100)severity: Filter by severity levels
list_scans
List all previous security scans.
Parameters:
limit: Maximum number of scans to returntarget: Filter by target
schedule_periodic_scan
Schedule recurring security scans (requires agent runtime).
Parameters:
targets(required): List of targets to scaninterval_hours: Scan interval (default: 24)severity_threshold: Minimum severity to report (default: "medium")notify_on_new: Alert on new vulnerabilities (default: true)
Integration with Agentic System
Cluster-Wide Scanning
The security scanner automatically detects and scans all nodes in the agentic cluster. Cluster nodes are loaded from configuration:
builder - Linux build node
orchestrator - Coordination node
coordinator - Multi-node coordinator
files - File server
Autonomous Agent Integration
Combine with the autonomous security scanning agent for:
Automated vulnerability assessment
Continuous security monitoring
Intelligent threat prioritization
Automatic remediation recommendations
Enhanced Memory Integration
Scan results are stored in enhanced-memory for:
Historical vulnerability tracking
Pattern recognition across scans
Causal relationship analysis
Learning from remediation outcomes
Scan Results
Results are stored in ${AGENTIC_SYSTEM_PATH:-/opt/agentic}/security-scans/:
scan_YYYYMMDD_HHMMSS_ID.json- Scan findingsscan_YYYYMMDD_HHMMSS_ID_metadata.json- Scan metadata
Example Usage
Scan a single target
# Via Claude Code
scan_target({
"target": "192.0.2.196",
"severity": ["high", "critical"],
"templates": ["network", "exposure"]
})Scan entire cluster
scan_cluster_nodes({
"scan_type": "full",
"severity": ["medium", "high", "critical"]
})Review previous scans
list_scans({"limit": 10})
get_scan_results({"scan_id": "scan_20251118_110000_1234"})Security Considerations
Scans generate network traffic - coordinate with network admin
Rate limiting prevents overwhelming targets
Results may contain sensitive information - restrict access
Authorized scanning only - verify permission before scanning external targets
Dependencies
Nuclei v3.5.1+
Python 3.10+
asyncio
aiofiles
pydantic
References
Nuclei - Main vulnerability scanner
nuclei-mcp - Reference MCP implementation
ExternalAttacker-MCP - Security testing MCP
License
MIT License - Part of the Mac Pro 5,1 Agentic System
Part of the MCP Ecosystem
This server integrates with other MCP servers for comprehensive AGI capabilities:
Server | Purpose |
4-tier persistent memory with semantic search | |
Persistent task queues and goal decomposition | |
Full AGI orchestration with 21 tools | |
Distributed task routing across nodes | |
Inter-node AI communication | |
Production-only policy enforcement |
See agentic-system-oss for the complete framework.
Available Tools
8 toolsdetect_anomalous_findingsA
Use TPU-accelerated embeddings to detect anomalous security findings.
Compares findings against a baseline (previous scans or expected patterns) to identify unusual or novel vulnerabilities that may need urgent attention.
Args: scan_id: Current scan to analyze baseline_scan_id: Previous scan to compare against (optional) threshold: Similarity threshold - lower values = more anomalies detected
Returns: JSON with anomalous findings and analysis
Inputs: scan_id (required), threshold, baseline_scan_id. Returns: result
get_scan_resultsA
Retrieve results from a previous scan.
Args: scan_id: Scan ID to retrieve limit: Maximum number of results to return (default: 100)
Returns: JSON string with scan results
Inputs: scan_id (required), limit. Returns: result
list_scansA
List all previous security scans.
Args: limit: Maximum number of scans to return (default: 50)
Returns: JSON string with scan list
Inputs: limit. Returns: result
list_templatesA
List available Nuclei templates by tag or severity.
Args: tag: Filter by tag (e.g., cve, exposure, misconfiguration) severity: Filter by severity (info, low, medium, high, critical)
Returns: JSON string with template list
Inputs: tag, severity. Returns: result
prioritize_findingsC
Use TPU-accelerated importance scoring to prioritize security findings.
Scores each finding based on semantic similarity to critical security terms and patterns, helping focus remediation efforts.
Args: scan_id: Scan ID to prioritize
Returns: JSON with prioritized findings
Inputs: scan_id (required). Returns: result
scan_cluster_nodesA
Scan all nodes in the agentic cluster for vulnerabilities.
Args: scan_type: Type of scan - "quick", "comprehensive", "full", "web", or "api" severity: Comma-separated severity levels to check
Returns: JSON string with cluster scan results
Inputs: severity, scan_type. Returns: result
scan_targetA
Scan a target URL or IP with Nuclei vulnerability scanner.
Args: target: Target URL or IP address to scan severity: Comma-separated severity levels (info, low, medium, high, critical) rate_limit: Requests per second (default: 150) timeout: Scan timeout in seconds (default: 300)
Returns: JSON string with scan results
Inputs: target (required), timeout, severity, rate_limit. Returns: result
update_templatesA
Update Nuclei templates to latest version.
Returns: JSON string with update result
Inputs: no named parameters. Returns: result
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 8 tools
Each tool has a clearly distinct purpose: scanning, results retrieval, template management, and analysis. No overlap exists.
All tool names follow a consistent verb_noun pattern in snake_case (e.g., scan_target, list_scans, detect_anomalous_findings). No mixing of conventions.
8 tools cover the domain well without being excessive. Each tool contributes meaningfully to scanning, results, templates, and analysis.
Core workflows are covered: scanning (targets, clusters), results retrieval, template management, anomaly detection, and prioritization. Minor gap: no tool to delete scans or clear data, but overall complete.
Maintenance
Related MCP Connectors
AI pentesting: run scans, triage vulnerabilities, review PRs, manage schedules and assets.
Autonomous AppSec & DAST scanner across 12 security domains (260+ controls) with 3-Layer Triage.
Live CVE, CISA KEV and exploit intelligence, SBOM scans, patch prioritization and IOC checks.
Scan a website for vulnerabilities: OWASP Top 10, CVEs, SSL, headers - with plain-English fixes
Related MCP Servers
- FlicenseNot gradedqualityDmaintenanceEnables vulnerability scanning of web applications using the xray security scanner. Provides URL scanning capabilities with configurable crawlers, plugins, and POCs to detect security vulnerabilities through natural language interactions.1-
- AlicenseNot gradedqualityDmaintenanceEnables network vulnerability scanning and CVE detection through an AI assistant, with optional remote port blocking via Discord bot.MIT
- AlicenseNot gradedqualityDmaintenanceEnables scanning IP addresses for vulnerabilities using Nmap and API-based checks, supporting both single and batch scans with detailed reports.6MIT
- AlicenseNot gradedqualityDmaintenanceEnables Claude Desktop to run real vulnerability scans using Nuclei through natural language commands.MIT