CrowdSentinel MCP Server provides comprehensive AI-driven threat hunting and incident response capabilities for Elasticsearch/OpenSearch, with a rich set of tools, extensive detection rules, and multi-source integration. However, its development status, security configuration defaults, and potential sensitive data access make it better suited for research and testing environments rather than production.
Strongest · Setup experience 14/20 Weakest · Reliability 8/20
Static review · not runListed 2026-08-07
Read the FMRS scoring method →Runtime：Python · Prebuilt binary
ELASTICSEARCH_API_KEY
Elasticsearch API key, recommended for production and Elastic Cloud; generate it in your cluster.
ELASTICSEARCH_PASSWORD
Basic auth password, used with ELASTICSEARCH_USERNAME.
ELASTICSEARCH_BEARER_TOKEN
Service/bearer token; highest auth priority, alternative to API key.
ELASTICSEARCH_CLIENT_KEY
Client private key path for mTLS.
VIRUSTOTAL_API_KEY
VirusTotal API key, free tier 500 lookups/day; optional.
ABUSEIPDB_API_KEY
AbuseIPDB API key, free tier 1,000 lookups/day; optional.
THREATFOX_API_KEY
abuse.ch ThreatFox API key, free unlimited; optional.
MISP_API_KEY
MISP API key (40-char hex).
ANTHROPIC_API_KEY
Anthropic auth for the CLI `analyse --mcp` agent mode; optional.
OPENAI_API_KEY
OpenAI auth for the CLI `analyse --mcp` agent mode; optional (browser login or local models also work).
ELASTICSEARCH_HOSTS
Comma-separated Elasticsearch hosts (e.g. https://localhost:9200); this is the default.
ELASTICSEARCH_CLOUD_ID
Elastic Cloud deployment ID, an alternative to ELASTICSEARCH_HOSTS for cloud deployments.
ELASTICSEARCH_USERNAME
Basic auth username, used with ELASTICSEARCH_PASSWORD.
VERIFY_CERTS
TLS verification: true (production), false (skip, dev only), or a CA cert path; default false.
REQUEST_TIMEOUT
Request timeout in seconds, e.g. 30 or 10.5.
ELASTICSEARCH_CA_CERT
Path to a CA certificate for self-signed setups.
ELASTICSEARCH_CLIENT_CERT
Client certificate path for mTLS.
DISABLE_HIGH_RISK_OPERATIONS
Set to true to hide the 10 high-risk write tools for read-only access.
MCP_ALLOWED_ORIGINS
Extra allowed Origins (comma-separated) for HTTP/SSE transports, anti DNS-rebinding.
VELOCIRAPTOR_API_CONFIG
Path to a Velociraptor API client YAML; setting it enables the 25 endpoint forensics tools.
MISP_URL
MISP instance URL for event export and attribute search; offline JSON export works without it.
MISP_SSL_VERIFY
Set to false for self-signed MISP instances in dev only.
Install via pip: pip install crowdsentinel-mcp-server. Run 'crowdsentinel setup' to download detection rules and tools. Set environment variables (ELASTICSEARCH_HOSTS, etc.) and run the server via uvx. Optional extras: velociraptor and deobf. For standard MCP clients use stdio transport; for web clients use SSE or HTTP.
{
"mcpServers": {
"crowdsentinel": {
"command": "uvx",
"args": [
"crowdsentinel-mcp-server"
],
"env": {
"ELASTICSEARCH_HOSTS": "https://localhost:9200",
"ELASTICSEARCH_API_KEY": "your_api_key",
"VERIFY_CERTS": "false"
}
}
}
}
Shown for Claude Code. Other clients may use a different file or key (VS Code uses "servers") — the configurator below converts it.
{
"servers": {
"crowdsentinel": {
"command": "uvx",
"args": [
"crowdsentinel-mcp-server"
],
"env": {
"ELASTICSEARCH_HOSTS": "https://localhost:9200",
"ELASTICSEARCH_API_KEY": "your_api_key",
"VERIFY_CERTS": "false"
}
}
}
}
Goes in your project's .vscode/mcp.json (VS Code uses a "servers" key).
claude mcp add crowdsentinel -e ELASTICSEARCH_HOSTS=https://localhost:9200 -e ELASTICSEARCH_API_KEY=your_api_key -e VERIFY_CERTS=false -- uvx crowdsentinel-mcp-server
Run it in a terminal; replace any <…> placeholders with your own values first.
After connecting, the client's tool list should show CrowdSentinel's 139 MCP tools (e.g. hunt_by_timeframe, get_host_activity_timeline); you can also run `crowdsentinel doctor` for an offline config check, or `crowdsentinel health` to confirm the Elasticsearch connection.
Once connected, you can ask your AI assistant things like:
Listed from the project's documentation, not tested by this site.
CrowdSentinel MCP Server is an open-source threat hunting orchestrator that connects large language models to enterprise security data via the Model Context Protocol (MCP). It provides 139 MCP tools for natural language threat hunting, detection rule execution (6,060 built-in rules), endpoint forensics (Velociraptor), host forensics (osquery), EVTX log analysis (Chainsaw), and PCAP network analysis (Wireshark). It supports multiple security frameworks (Cyber Kill Chain, Pyramid of Pain, Diamond Model, MITRE ATT&CK) and features persistent investigation state, cross-tool IoC correlation, and encoded-command deobfuscation. The server is intended for security testing, research, and educational purposes only, and is not production-ready.
Connect ClickHouse to your AI assistants
Unofficial MCP server providing advanced access to the ChEMBL chemical database
Google's official database MCP toolbox — define AI-accessible database tools via config
Bytebase's official, minimalist two-tool database MCP server
Source revision bfca612f4672 Data synced 2026-10-19 Read the FMRS scoring method