Intelthreadlinqs MCP README

Security Detections MCP

Unified Detection Engineering Platform
7,283+ detections across Sigma, Splunk ESCU, Elastic & KQL — powered by MCP

Quick Start • Workbench • 106 Tools • 22 Workflows • Agents • Config

Built on Security Detections MCP by MHaggis • Threat Intelligence by Threadlinqs

Dashboard

What is this?

Security Detections MCP is an open-source platform that unifies 7,283+ detection rules from four major sources into a single queryable interface using the Model Context Protocol (MCP). It combines:

An MCP Server with 74 tools for searching, analyzing, and engineering detections
A Threat Intelligence MCP Server (Threadlinqs Intel) with 32 tools for real-time threat feeds, C2 tracking, MITRE mapping, and DNS enrichment
A Web Dashboard (Detection Workbench) for visual exploration, workflow execution, and AI-powered ad-hoc analysis
An Autonomous Agent Pipeline (LangGraph) for end-to-end detection generation and validation

Detection Sources

Source	Rules	Format	Description
Sigma	3,200+	YAML	Community-driven, platform-agnostic detection rules
Splunk ESCU	1,800+	YAML/SPL	Splunk's Enterprise Security Content Update detections
Elastic	1,400+	TOML/EQL	Elastic Security detection rules
KQL	880+	KQL	Microsoft Sentinel / Defender analytics rules

Quick Start

Prerequisites

Node.js >= 18.0.0
npm >= 8.0.0
Git (for cloning detection repos)

1. Clone and build the MCP server

bash

git clone https://github.com/MHaggis/Security-Detections-MCP.git
cd Security-Detections-MCP
npm install
npm run build

2. Clone detection rule repositories

bash

mkdir -p detections && cd detections

# Sigma rules
git clone https://github.com/SigmaHQ/sigma.git

# Splunk ESCU
git clone https://github.com/splunk/security_content.git

# Elastic detection rules
git clone https://github.com/elastic/detection-rules.git

# KQL rules
git clone https://github.com/bert-jan/kql.git kql-bertjanp
git clone https://github.com/jkerai1/SentinelAnalyticRules.git kql-jkerai1

cd ..

3. Start the workbench

bash

cd workbench
npm install
npm run dev

Open http://localhost:3000 in your browser.

4. Configure API keys

Navigate to Settings in the workbench sidebar and configure:

LLM Provider — Choose Anthropic (Claude) or OpenAI (GPT-4) and enter your API key
Threadlinqs Intel API Key — Get yours at intel.threadlinqs.com for real-time threat intelligence

Without an LLM API key, ad-hoc queries and llm_prompt workflow steps won't function. Without a Threadlinqs key, threat intelligence workflows will skip intel steps but detection tools will work normally.

Architecture

Code

┌─────────────────────────────────────────────────────────────────┐
│                    Detection Workbench (Next.js)                │
│  Dashboard │ Tools │ Workflows │ Ad-Hoc │ Threats │ Coverage    │
├──────────────────────┬──────────────────────────────────────────┤
│    MCP Bridge        │         LLM Client                      │
│  (StdioTransport)    │   (Anthropic / OpenAI)                  │
├──────────┬───────────┼──────────────────────────────────────────┤
│          │           │                                          │
│  ┌───────▼────────┐  │  ┌─────────────────────┐                │
│  │ security-      │  │  │ threadlinqs-intel    │                │
│  │ detections     │  │  │ MCP Server (v3.1.1)  │                │
│  │ MCP Server     │  │  │                      │                │
│  │                │  │  │ 32 tools:            │                │
│  │ 74 tools:      │  │  │ • Threat feeds       │                │
│  │ • Search       │  │  │ • C2 tracking        │                │
│  │ • MITRE map    │  │  │ • MITRE mapping      │                │
│  │ • Knowledge    │  │  │ • CVE/CWE lookup     │                │
│  │ • Engineering  │  │  │ • IOC search         │                │
│  │ • Autonomous   │  │  │ • Simulations        │                │
│  └───────┬────────┘  │  └──────────┬──────────┘                │
│          │           │             │                            │
│  ┌───────▼───────────▼─────────────▼──────────┐                │
│  │              SQLite Cache                   │                │
│  │         7,283+ indexed detections           │                │
│  └───────────────┬────────────────────────────┘                │
│                  │                                              │
│  ┌───────────────▼────────────────────────────┐                │
│  │         Detection Rule Repositories         │                │
│  │   Sigma │ Splunk ESCU │ Elastic │ KQL       │                │
│  └─────────────────────────────────────────────┘                │
└─────────────────────────────────────────────────────────────────┘

Detection Workbench

The workbench is a full-featured web dashboard built with Next.js 15, React 19, and Tailwind CSS v4. It features a terminal-inspired hacker aesthetic with JetBrains Mono font, colored status indicators, and // section headers.

Dashboard

The command center with real-time stats, quick-launch workflows, recent activity, and server status at a glance.

Dashboard

Metric	Value
Total Tools	106
Indexed Detections	7,283+
Pre-built Workflows	22
MCP Servers	2

Tool Explorer

Browse, search, and execute all 106 tools across both MCP servers. Blue indicators for security-detections, purple for threadlinqs-intel.

Tool Explorer

Click any tool to inspect its schema, parameters, and test it directly from the browser.

Filter by server to see the 32 Threadlinqs Intel tools — threat feeds, C2 tracking, simulations, and more.

Intel Tools

Workflow Library

22 pre-built detection engineering workflows spanning threat hunting, coverage analysis, CVE assessment, and intelligence operations.

Workflows

Visual Workflow Builder

Drag-and-drop canvas for designing custom detection pipelines. Connect tool calls, transforms, LLM analysis steps, conditionals, loops, and parallel branches.

Workflow Builder

Workflow Execution

Execute workflows against live MCP servers with real-time step-by-step progress tracking. Each step shows its type (tool_call, transform, llm_prompt), target server, and execution status.

Workflow Execution

Results include LLM-generated analysis with executive summaries, threat inventories, MITRE coverage matrices, and prioritized recommendations.

Execution Results

Ad-Hoc Terminal

Natural language interface powered by your configured LLM. Ask questions about your detections, threats, and coverage — the system automatically selects and calls the right MCP tools, then synthesizes the results.

Ad-Hoc Terminal

Example queries:

"What ransomware detections do we have?"
"Show coverage gaps for credential access"
"Compare Sigma vs Splunk detection counts"
"Pull the latest threats from Threadlinqs and contrast them against my detections"

Ad-Hoc Results

All Pages

Page	Path	Description
Dashboard	`/`	Stats, quick-launch, recent activity, server status
Ad-Hoc Terminal	`/adhoc`	Natural language queries with automatic tool orchestration
Tool Explorer	`/tools`	Browse and execute all 106 MCP tools
Workflow Library	`/workflows`	View and run 22 pre-built workflows
Workflow Builder	`/workflows/builder`	Visual drag-and-drop pipeline designer
Threats	`/threats`	Real-time threat intelligence feed
Coverage	`/coverage`	MITRE ATT&CK coverage analysis
History	`/history`	Execution history and past results
Settings	`/config`	LLM provider, API keys, MCP server status

MCP Tools

security-detections (74 tools)

Core Detection Tools

Tool	Params	Description
`search`	query, limit, source_type	Full-text search across all detection fields
`get_by_id`	id	Get detection by unique identifier
`get_raw_yaml`	id	Get original YAML/TOML source
`list_all`	limit, offset	List all detections with pagination
`list_by_source`	source, limit, offset	Filter by source (sigma, splunk_escu, elastic, kql)
`list_by_mitre`	technique_id, limit	Filter by MITRE ATT&CK technique
`list_by_logsource`	category, product, service, limit, offset	Filter by log source fields
`list_by_severity`	severity, limit, offset	Filter by severity level
`list_by_cve`	cve_id, limit	Find detections related to a CVE
`list_by_analytic_story`	story, limit	Filter by Splunk analytic story
`list_by_process_name`	process, limit, offset	Filter by process name
`list_by_detection_type`	type, limit, offset	Filter by detection type
`list_by_data_source`	data_source, limit, offset	Filter by data source
`list_by_mitre_tactic`	tactic, limit, offset	Filter by MITRE tactic
`list_by_kql_category`	category, limit, offset	Filter KQL by category
`list_by_kql_tag`	tag, limit, offset	Filter KQL by tag
`list_by_kql_datasource`	datasource, limit, offset	Filter KQL by data source
`list_by_name_pattern`	pattern, limit, offset	Wildcard search on detection name

Analytics & Coverage

Tool	Description
`get_stats`	Platform-wide detection statistics
`get_technique_ids`	All MITRE technique IDs with detection counts
`get_technique_count`	Count detections for a specific technique
`analyze_coverage`	Analyze MITRE coverage with gap identification
`identify_gaps`	Find techniques with missing or thin coverage
`suggest_detections`	Get recommendations for new detections to build
`get_detection_list`	Generate filterable detection inventory
`compare_sources`	Compare coverage across Sigma, Splunk, Elastic, KQL
`count_by_source`	Count detections per source type
`smart_compare`	Intelligent cross-source comparison
`get_coverage_summary`	Executive-level coverage summary
`get_top_gaps`	Highest-priority detection gaps

Analytic Stories

Tool	Description
`search_stories`	Full-text search across analytic stories
`get_story`	Get detailed story information
`list_stories`	List all analytic stories
`list_stories_by_category`	Filter stories by category

Knowledge Graph (Tribal Knowledge)

Tool	Description
`create_entity`	Create knowledge entities (threat actors, techniques, etc.)
`create_relation`	Create relationships between entities with reasoning
`add_observation`	Add facts with confidence scoring
`delete_entity`	Remove entity and associated data
`delete_observation`	Delete specific observation
`search_knowledge`	Full-text search across all knowledge
`read_graph`	Read entire graph or filtered subgraph
`open_entity`	Full entity details with relations and observations
`log_decision`	Record significant decisions with reasoning
`add_learning`	Store reusable patterns and insights
`get_relevant_decisions`	Find past decisions relevant to current context
`get_learnings`	Get applicable learnings for current task

Engineering Intelligence

Tool	Description
`get_query_patterns`	Common query patterns for a MITRE technique
`get_field_reference`	Available fields for a Splunk data model
`get_macro_reference`	Common Splunk macros and usage patterns
`find_similar_detections`	Find detections similar to what you want to create
`suggest_detection_template`	Generate detection template from technique + patterns

Dynamic Tables

Tool	Description
`create_table`	Create custom runtime tables
`insert_row`	Insert data into dynamic table
`query_table`	Query with filtering, sorting, pagination
`list_tables`	List all tables (including pre-built analysis tables)
`drop_table`	Remove table and data
`describe_table`	Get schema and statistics

Pre-built tables: gap_analyses, source_comparisons, threat_actor_profiles, detection_recommendations

Autonomous Analysis

Tool	Description
`auto_analyze_coverage`	Automatic coverage analysis across threat profiles
`auto_gap_report`	Comprehensive gap report with prioritized recommendations
`auto_compare_sources`	Detailed cross-source comparison with reasoning
`llm_enhanced_analysis`	LLM-enhanced analysis via MCP sampling
`check_sampling_status`	Check if MCP sampling is available

Cache & Templates

Tool	Description
`save_query`	Cache query results with optional TTL
`get_saved_query`	Retrieve cached results
`list_saved_queries`	List all saved queries
`rebuild_index`	Force re-index all detections
`save_template`	Save reusable query templates with `{{placeholders}}`
`run_template`	Execute template with parameter substitution
`list_templates`	List all templates with usage stats
`get_template`	Get template details and schema
`delete_template`	Remove a template

threadlinqs-intel (32 tools) — `intelthreadlinqs-mcp@3.1.1`

Threat Intelligence

Tool	Description
`search_threats`	Search threat intelligence database by keyword, CVE, MITRE technique, or actor
`get_threat`	Get full threat report with TTPs, IOCs, detections, and attribution
`get_recent_threats`	Get most recently published threats, ordered by creation date
`list_threat_categories`	List all threat categories with counts
`get_threat_transcripts`	Multi-agent intelligence analysis transcripts for a threat

Detections & Coverage

Tool	Description
`get_detections`	Get detection rules, filter by threat ID or type (SPL, KQL, Sigma)
`export_detection`	Export detection in raw SIEM format
`get_mitre_coverage`	Full MITRE ATT&CK framework coverage map
`get_mitre_technique`	Detailed info for a specific MITRE ATT&CK technique

IOC Intelligence & DNS Enrichment

Tool	Description
`search_iocs`	Search IOCs — IPs, domains, hashes, URLs, behavioral indicators
`enrich_iocs`	Trigger DNS enrichment for IOCs (single, per-threat, or bulk)
`get_ioc_dns`	Get passive DNS enrichment data for a specific IOC (IP or domain)
`get_enrichment_graph`	DNS enrichment infrastructure graph — nodes and edges
`get_enrichment_overview`	DNS enrichment overview — totals, discovered domains, cross-threat links

C2 Tracking

Tool	Description
`list_c2_beacons`	Wild C2 beacons — command-and-control infrastructure intel
`get_c2_stats`	C2 framework distribution, geographic spread, watermarks
`get_c2_operators`	C2 operator cluster analysis — behavioral fingerprints
`get_c2_cross_correlations`	Cross-intelligence correlations — IP, tag, MITRE matches
`get_c2_watermarks`	Cobalt Strike watermark clusters
`get_c2_timeline`	Monthly C2 framework version distribution
`get_c2_configs`	Full C2 beacon configurations — sleep, jitter, user agents

Correlations & Lookups

Tool	Description
`get_correlations_overview`	Aggregated intelligence from 7 correlation engines
`get_correlation_engine`	Detailed data from a specific correlation engine
`get_cve_details`	CVE lookup with severity, CVSS, and description
`get_cwe_details`	CWE weakness lookup with description and mitigations

Simulations & Debriefs

Tool	Description
`list_simulations`	List attack simulation scenarios
`get_threat_simulations`	Get simulation commands for a specific threat
`list_debriefs`	List daily intelligence debriefs
`get_debrief`	Get specific daily debrief by date

Platform

Tool	Description
`get_platform_stats`	Platform statistics — threats, detections, IOCs, coverage
`get_roadmap`	Platform roadmap — planned features and ideas
`get_changelog`	Recent updates, features, and fixes

Workflow Engine

The workflow engine supports 6 step types that can be chained into multi-step detection engineering pipelines:

Step Type	Description	Example
`tool_call`	Execute an MCP tool on a specified server	`search` on `security-detections`
`transform`	Extract or reshape data between steps	`$steps.threats[0].id`
`llm_prompt`	Send a prompt to the configured LLM for analysis	Coverage gap analysis
`conditional`	Branch execution based on a condition	`$steps.count > 0`
`loop`	Iterate over an array, executing steps per item	Loop over techniques
`parallel`	Execute multiple steps concurrently	Query multiple sources

Binding System

Steps can reference outputs from previous steps using $-expressions:

Code

$input.cve_id              # Workflow input parameter
$steps.threats             # Output from step with id "threats"
$steps.threats[0].id       # First item's id field
$steps.detail.mitre_attack # Nested field access

Pre-built Workflows (22)

Threat Intelligence

Workflow	Tags	Description
Threat Coverage Check	`threats, coverage, gaps`	Get recent threats, extract MITRE techniques, identify detection gaps
CVE Impact Assessment	`cve, threats, assessment`	Assess CVE impact by finding related threats and detections
Daily Threat Debrief	`daily, debrief, threats`	Generate daily threat intelligence summary with coverage report
New Threat Triage	`triage, threat, detections`	Triage specific threat with detection check and recommendations
Threat Actor Profile	`threat-actor, profile, techniques`	Build complete threat actor profile with TTPs and coverage
IOC Enrichment	`ioc, enrichment, threats`	Enrich IOCs with threat context, technique mapping, coverage check
Weekly Security Report	`weekly, report, summary`	Comprehensive weekly report with stats and correlations

Coverage Analysis

Workflow	Tags	Description
Detection Gap Analysis	`gaps, coverage, mitre`	Compare MITRE coverage between threat intel and detections
Coverage Sync	`sync, coverage, mitre`	Parallel sync of MITRE coverage across intel and detections
Detection Source Comparison	`comparison, sources`	Compare counts and coverage across Sigma, Splunk, Elastic, KQL
Detection Quality Report	`quality, optimization`	Find thin coverage, over-covered areas, rebalancing suggestions

Threat Hunting

Workflow	Tags	Description
C2 Framework Hunting	`c2, hunting`	Identify C2 frameworks and check detection coverage
Lateral Movement Hunt	`lateral-movement, hunting`	Map lateral movement techniques, build hunting plan
Credential Theft Audit	`credentials, audit`	Audit credential dumping, brute force, kerberoasting detections
Exfiltration Watchlist	`exfiltration, dlp`	Map exfiltration techniques, C2 channels, DLP blind spots

Posture Assessment

Workflow	Tags	Description
Ransomware Readiness	`ransomware, readiness`	Evaluate detection posture against ransomware TTPs
Initial Access Posture	`initial-access, phishing`	Assess phishing, drive-by, supply chain, valid account coverage
Privilege Escalation Review	`privesc, escalation`	Deep dive into UAC bypass, token manipulation, process injection
PowerShell Threat Matrix	`powershell, lolbas`	PowerShell attack analysis: obfuscation, AMSI bypass, LOLBins
Cloud Security Coverage	`cloud, aws, azure, gcp`	Cloud-specific attack coverage across AWS, Azure, GCP
Defense Evasion Deep Dive	`evasion, defense`	Exhaustive analysis of the largest MITRE tactic
Threat Simulation Readiness	`simulation, purple-team`	Check if detection stack catches simulated TTPs

Autonomous Pipeline

The agents/ directory contains a LangGraph-based autonomous pipeline for end-to-end detection generation and validation:

Code

CTI Analyst ──► Coverage Analyzer ──► Detection Engineer ──► Atomic Executor
                                                                    │
                                                            Splunk Validator
                                                                    │
                                                              Data Dumper
                                                                    │
                                                              PR Stager

Agent Nodes

Agent	Role
CTI Analyst	Analyzes threat intelligence and extracts TTPs
Coverage Analyzer	Maps existing detection coverage against threats
Detection Engineer	Generates new detection rules for gaps
Atomic Executor	Creates Atomic Red Team test cases
Splunk Validator	Validates SPL syntax and logic
SIEM Validator	Cross-platform SIEM validation
FP Analyst	Assesses false positive risk
QA Reviewer	Quality assurance review
Verifier	End-to-end verification
Data Dumper	Exports results to structured formats
PR Stager	Stages detection PRs for review
Attack Range Builder	Provisions test environments

Setup

bash

cd agents
cp .env.example .env
# Edit .env with your ANTHROPIC_API_KEY and SIEM_PLATFORM
npm install
npm run build

Configuration

MCP Server Configuration

The MCP servers are configured in .cursor/mcp.json:

config.json

{
  "mcpServers": {
    "security-detections": {
      "command": "node",
      "args": ["./dist/index.js"],
      "env": {
        "SIGMA_PATHS": "./detections/sigma/rules,./detections/sigma/rules-threat-hunting",
        "SPLUNK_PATHS": "./detections/security_content/detections",
        "ELASTIC_PATHS": "./detections/detection-rules/rules",
        "KQL_PATHS": "./detections/kql-bertjanp,./detections/kql-jkerai1",
        "STORY_PATHS": "./detections/security_content/stories"
      }
    },
    "threadlinqs-intel": {
      "command": "npx",
      "args": ["-y", "intelthreadlinqs-mcp@3.1.1"],
      "env": {
        "THREADLINQS_API_KEY": ""
      }
    }
  }
}

LLM Configuration

The workbench supports two LLM providers configured in the Settings page:

Provider	Models	API Key Env Var
Anthropic	`claude-sonnet-4-20250514`, `claude-haiku-4-20250414`, `claude-opus-4-20250514`	`ANTHROPIC_API_KEY`
OpenAI	`gpt-4o`, `gpt-4o-mini`, `gpt-4-turbo`	`OPENAI_API_KEY`

API keys can be set in three ways (in priority order):

Settings page in the workbench UI
Environment variables (ANTHROPIC_API_KEY, OPENAI_API_KEY, THREADLINQS_API_KEY)
.env.local file in the workbench/ directory

Workbench Engine Settings

Setting	Default	Description
`max_prompt_size`	12,000 chars	Maximum prompt length sent to LLM
`per_binding_limit`	3,000 chars	Max chars per step binding in prompts
`llm_timeout`	180s	LLM request timeout
`error_policy`	`continue`	`continue` or `stop` on step failure
`max_tokens`	4,096	Maximum LLM response tokens

Using with AI Assistants

Claude Code (CLI)

Add the MCP servers to your Claude Code configuration:

bash

# Add the security-detections MCP server
claude mcp add security-detections \
  -e SIGMA_PATHS="/path/to/detections/sigma/rules,/path/to/detections/sigma/rules-threat-hunting" \
  -e SPLUNK_PATHS="/path/to/detections/security_content/detections" \
  -e ELASTIC_PATHS="/path/to/detections/detection-rules/rules" \
  -e KQL_PATHS="/path/to/detections/kql-bertjanp,/path/to/detections/kql-jkerai1" \
  -e STORY_PATHS="/path/to/detections/security_content/stories" \
  -- node /path/to/Security-Detections-MCP/dist/index.js

# Add the threadlinqs-intel MCP server
claude mcp add threadlinqs-intel \
  -e THREADLINQS_API_KEY="your-threadlinqs-api-key" \
  -- npx -y intelthreadlinqs-mcp@3.1.1

Or add them directly to ~/.claude.json (global) or .claude/settings.json (project-level):

config.json

{
  "mcpServers": {
    "security-detections": {
      "command": "node",
      "args": ["/path/to/Security-Detections-MCP/dist/index.js"],
      "env": {
        "SIGMA_PATHS": "/path/to/detections/sigma/rules,/path/to/detections/sigma/rules-threat-hunting",
        "SPLUNK_PATHS": "/path/to/detections/security_content/detections",
        "ELASTIC_PATHS": "/path/to/detections/detection-rules/rules",
        "KQL_PATHS": "/path/to/detections/kql-bertjanp,/path/to/detections/kql-jkerai1",
        "STORY_PATHS": "/path/to/detections/security_content/stories"
      }
    },
    "threadlinqs-intel": {
      "command": "npx",
      "args": ["-y", "intelthreadlinqs-mcp@3.1.1"],
      "env": {
        "THREADLINQS_API_KEY": "your-threadlinqs-api-key"
      }
    }
  }
}

Claude Desktop / Cursor / Windsurf

Add the servers to your MCP configuration file (.cursor/mcp.json, claude_desktop_config.json, etc.):

config.json

{
  "mcpServers": {
    "security-detections": {
      "command": "node",
      "args": ["/path/to/Security-Detections-MCP/dist/index.js"],
      "env": {
        "SIGMA_PATHS": "/path/to/detections/sigma/rules",
        "SPLUNK_PATHS": "/path/to/detections/security_content/detections",
        "ELASTIC_PATHS": "/path/to/detections/detection-rules/rules",
        "KQL_PATHS": "/path/to/detections/kql-bertjanp,/path/to/detections/kql-jkerai1"
      }
    },
    "threadlinqs-intel": {
      "command": "npx",
      "args": ["-y", "intelthreadlinqs-mcp@3.1.1"],
      "env": {
        "THREADLINQS_API_KEY": "your-threadlinqs-api-key"
      }
    }
  }
}

Example Prompts

Code

"Search for all detections related to T1059.001 (PowerShell)"

"Compare Sigma vs Splunk ESCU coverage for credential access techniques"

"Find detection gaps in our lateral movement coverage"

"What detections do we have for ransomware encryption behaviors?"

"Create a knowledge entity for APT29 with all known techniques"

"Generate a gap report for the top 10 most critical MITRE techniques"

API Reference

The workbench exposes a REST API for programmatic access:

Method	Endpoint	Description
`GET`	`/api/mcp/servers`	List MCP servers and status (auto-connects on first call)
`POST`	`/api/mcp/servers`	Reconnect a server (`{action: "reconnect", server: "name"}`)
`GET`	`/api/mcp/tools`	List all available tools
`POST`	`/api/mcp/{server}/{tool}`	Execute a tool
`GET`	`/api/workflows`	List all workflows
`POST`	`/api/workflows`	Create a workflow
`GET`	`/api/workflows/{id}`	Get workflow details
`PUT`	`/api/workflows/{id}`	Update a workflow
`POST`	`/api/workflows/{id}/execute`	Execute a workflow
`GET`	`/api/workflows/{id}/stream`	Stream execution events (SSE)
`GET`	`/api/executions`	List execution history
`DELETE`	`/api/executions`	Clear execution history
`POST`	`/api/adhoc`	Execute ad-hoc query (SSE stream)
`GET`	`/api/settings`	Get settings (keys masked)
`POST`	`/api/settings`	Update settings
`GET`	`/api/settings?test=true`	Test LLM connection

Project Structure

Code

Security-Detections-MCP/
├── src/                          # MCP server source
│   ├── index.ts                  # Server entry point
│   ├── db/                       # SQLite database layer
│   ├── handlers/                 # MCP protocol handlers
│   ├── resources/                # MCP resources
│   ├── tools/                    # 74 MCP tools
│   │   ├── detections/           # Core detection query tools
│   │   ├── stories/              # Analytic story tools
│   │   ├── knowledge/            # Knowledge graph (12 tools)
│   │   ├── engineering/          # Detection engineering intelligence
│   │   ├── autonomous/           # Autonomous analysis tools
│   │   ├── dynamic/              # Dynamic table tools
│   │   ├── cache/                # Query caching tools
│   │   ├── meta/                 # Template tools
│   │   └── registry.ts           # Plugin-style tool registry
│   ├── types/                    # TypeScript type definitions
│   └── utils/                    # Shared utilities
│
├── workbench/                    # Detection Workbench (Next.js 15)
│   ├── app/                      # App Router pages & API routes
│   │   ├── api/                  # REST API endpoints
│   │   ├── adhoc/                # Ad-hoc terminal page
│   │   ├── tools/                # Tool explorer page
│   │   ├── workflows/            # Workflow library + builder
│   │   ├── threats/              # Threat intelligence page
│   │   ├── coverage/             # Coverage analysis page
│   │   ├── history/              # Execution history page
│   │   └── config/               # Settings page
│   ├── components/               # React components
│   │   ├── layout/               # Sidebar, header
│   │   ├── tools/                # Tool list, tool form
│   │   ├── shared/               # Loading, markdown renderer
│   │   └── workflow-builder/     # Visual workflow builder
│   ├── lib/                      # Core libraries
│   │   ├── mcp-pool.ts           # MCP server process pool
│   │   ├── mcp-config.ts         # MCP configuration loader
│   │   ├── llm-client.ts         # LLM abstraction (Anthropic/OpenAI)
│   │   ├── workflow-engine.ts    # Workflow execution engine
│   │   ├── workflow-presets.ts   # 22 pre-built workflows
│   │   └── transforms.ts        # $-binding expression resolver
│   └── types/                    # Workbench type definitions
│
├── agents/                       # Autonomous LangGraph pipeline
│   ├── nodes/                    # 12 agent node implementations
│   ├── graphs/                   # Orchestration graph definitions
│   ├── state/                    # LangGraph state management
│   ├── tools/                    # Specialized agent tools
│   └── tests/                    # Agent test suite
│
├── detections/                   # Detection rule repositories (git cloned)
│   ├── sigma/                    # SigmaHQ rules
│   ├── security_content/         # Splunk ESCU
│   ├── detection-rules/          # Elastic rules
│   ├── kql-bertjanp/             # KQL rules (bert-jan)
│   └── kql-jkerai1/              # KQL rules (jkerai1)
│
├── .cursor/
│   ├── mcp.json                  # MCP server configuration
│   ├── rules/                    # 6 Cursor AI rules
│   └── agents/                   # 15 Cursor agent definitions
│
├── docs/                         # Documentation & screenshots
└── package.json                  # Root package (v3.0.0)

Tech Stack

Layer	Technology
MCP Server	Node.js, TypeScript, `@modelcontextprotocol/sdk`, `better-sqlite3`
Workbench	Next.js 15, React 19, Tailwind CSS v4, Zustand, React Flow
LLM Integration	Anthropic Claude API, OpenAI API (configurable)
Threat Intel	Threadlinqs Intel MCP (`intelthreadlinqs-mcp@3.1.1`)
Agent Pipeline	LangGraph, TypeScript
Database	SQLite with FTS5 full-text search
Detection Formats	Sigma YAML, Splunk YAML/SPL, Elastic TOML/EQL, KQL

Development

Build the MCP server

Terminal

npm run build          # Compile TypeScript
npm run dev            # Watch mode
npm run lint           # Type check (strict)

Run the workbench

bash

cd workbench
npm run dev            # Development server (hot reload)
npm run build          # Production build
npm run start          # Production server

Run tests

Terminal

npm test               # Full test suite
npm run test:platform  # Cross-platform tests
npm run test:ci        # CI integration tests
npm run test:engineering  # Engineering tools tests

Contributing

Fork the repository
Create a feature branch (git checkout -b feature/amazing-detection)
Commit your changes (git commit -m 'Add amazing detection workflow')
Push to the branch (git push origin feature/amazing-detection)
Open a Pull Request

Acknowledgments

SigmaHQ for the Sigma detection rule standard
Splunk for Enterprise Security Content Update
Elastic for Elastic Security detection rules
Threadlinqs for real-time threat intelligence
Model Context Protocol for the MCP standard
MITRE ATT&CK for the adversary tactics and techniques framework

Built with MCP • Open Source • Detection Engineering

Security Detections MCP

What is this?

Detection Sources

Quick Start

Prerequisites

1. Clone and build the MCP server

2. Clone detection rule repositories

3. Start the workbench

4. Configure API keys

Architecture

Detection Workbench

Dashboard

Tool Explorer

Workflow Library

Visual Workflow Builder

Workflow Execution

Ad-Hoc Terminal

All Pages

MCP Tools

security-detections (74 tools)

Core Detection Tools

Analytics & Coverage

Analytic Stories

Knowledge Graph (Tribal Knowledge)

Engineering Intelligence

Dynamic Tables

Autonomous Analysis

Cache & Templates

threadlinqs-intel (32 tools) — intelthreadlinqs-mcp@3.1.1

Threat Intelligence

Detections & Coverage

IOC Intelligence & DNS Enrichment

C2 Tracking

Correlations & Lookups

Simulations & Debriefs

Platform

Workflow Engine

Binding System

Pre-built Workflows (22)

Threat Intelligence

Coverage Analysis

Threat Hunting

Posture Assessment

Autonomous Pipeline

Agent Nodes

Setup

Configuration

MCP Server Configuration

LLM Configuration

Workbench Engine Settings

Using with AI Assistants

Claude Code (CLI)

Claude Desktop / Cursor / Windsurf

Example Prompts

API Reference

Project Structure

Tech Stack

Development

Build the MCP server

Run the workbench

Run tests

Contributing

Acknowledgments

threadlinqs-intel (32 tools) — `intelthreadlinqs-mcp@3.1.1`