Skip to main content
Glama
UTMStack MCP Server

UTMStack MCP Server

Official
by utmstack

UTMStack MCP Server

A Model Context Protocol server that lets an AI assistant — Claude Code, OpenCode, Claude Desktop, or any MCP client — drive your UTMStack SIEM/XDR platform: triage alerts, search logs, run SQL, manage incidents, inspect agents, create and delete correlation rules, manage data filters, and run commands on endpoints.

Ships as a single self-contained binary. No Python installation required.


Install

Find your platform below and copy the matching command. Each installer detects your CPU architecture automatically — one command per operating system, not per architecture.

macOS — Apple Silicon (M1–M4) and Intel

curl -fsSL https://raw.githubusercontent.com/utmstack/MCP/main/install.sh | sh

Linux — x86_64 and ARM64

curl -fsSL https://raw.githubusercontent.com/utmstack/MCP/main/install.sh | sh

Windows — x64 and ARM64 (PowerShell)

irm https://raw.githubusercontent.com/utmstack/MCP/main/install.ps1 | iex

Run this in PowerShell, not cmd.exe, and not Git Bash — Git Bash cannot install the Windows build, and the macOS/Linux installer detects it and sends you back here. Inside WSL, use the macOS/Linux command instead: WSL needs the Linux build, which is what that command installs.

Supported platforms

Operating system

Architecture

Asset

macOS 12+

Apple Silicon (arm64)

utmstack-mcp-darwin-arm64.tar.gz

macOS 12+

Intel (x86_64)

utmstack-mcp-darwin-amd64.tar.gz

Linux (glibc 2.35+)

x86_64

utmstack-mcp-linux-amd64.tar.gz

Linux (glibc 2.39+)

arm64 / aarch64

utmstack-mcp-linux-arm64.tar.gz

Windows 10/11

x64

utmstack-mcp-windows-amd64.zip

Windows 11

ARM64

utmstack-mcp-windows-arm64.zip

Linux x86_64 targets glibc 2.35 (Ubuntu 22.04+, Debian 12+, RHEL 9+). Linux arm64 is built on a newer runner and targets glibc 2.39 (Ubuntu 24.04+, Debian 13+, Kali rolling). Alpine/musl is not covered — run from source there. 32-bit systems are not supported.

Every installer verifies the download's SHA256 against the checksums.txt published with the release, and refuses to install if it cannot.

Where it lands

Location

macOS / Linux

/usr/local/bin/utmstack-mcp, or ~/.local/bin without sudo

Windows

%LOCALAPPDATA%\Programs\utmstack-mcp\utmstack-mcp.exe

Set UTMSTACK_MCP_VERSION=1.2.3 to install a specific version.

Unsigned binaries

These binaries are not code-signed. They run normally — signing is not required to execute anything on macOS or Windows — but it affects two situations:

macOS. Binaries fetched by curl are not quarantined, so the installer path works as-is. If you download an asset from the Releases page in a browser, Gatekeeper will block it — use the installer, or clear the flag:

xattr -dr com.apple.quarantine /usr/local/lib/utmstack-mcp

Windows. SmartScreen triggers on Mark-of-the-Web, which browsers apply to downloads; Invoke-WebRequest (what install.ps1 uses) does not, so the installer path is normally silent. Separately, Microsoft Defender is known to heuristically flag PyInstaller-built executables. If it quarantines utmstack-mcp.exe, the file is a false positive — verify its SHA256 against checksums.txt on the release, then add an exclusion for %LOCALAPPDATA%\Programs\utmstack-mcp.

Every release publishes checksums.txt, and the installers verify against it and refuse to proceed if they cannot.


Related MCP server: AlertLogic MCP Server

Set up

utmstack-mcp init

The wizard asks for your UTMStack URL, how to authenticate, and your TLS preference, then validates the credentials against the live API before saving anything. It writes them to a 0600 file (~/.config/utmstack-mcp/config.json, or %APPDATA%\utmstack-mcp on Windows), so no credentials end up in shell history or in an MCP client's config.

Check it any time:

utmstack-mcp check        # verify configuration and connectivity
utmstack-mcp config-path  # print the config file location

Connect an MCP client

Because credentials live in the config file, registration carries no secrets and is the same everywhere.

Claude Code

claude mcp add --scope user utmstack -- utmstack-mcp

OpenCode — add to ~/.config/opencode/opencode.json:

{ "mcp": { "utmstack": { "type": "local", "command": ["utmstack-mcp"], "enabled": true } } }

Claude Desktop — add to claude_desktop_config.json:

{ "mcpServers": { "utmstack": { "command": "utmstack-mcp" } } }

The UTMStack CLI bundles this server preconfigured — nothing to register there.


Configuration

utmstack-mcp init writes this file for you. Configuration is resolved in this order, first match wins:

  1. UTMSTACK_SERVERS (a JSON array) in the environment

  2. UTMSTACK_SERVERS_FILE pointing at a JSON file

  3. ~/.config/utmstack-mcp/config.json — what init writes

  4. .env in the current directory (development convenience)

[
  {
    "name": "prod",
    "url": "https://utm.example.com",
    "user": "admin",
    "pass": "...",
    "default": true
  },
  {
    "name": "lab",
    "url": "https://10.0.0.2",
    "apiKey": "...",
    "caBundle": "/etc/ssl/certs/lab-ca.pem",
    "allowAgentCommands": true
  }
]

Key

Meaning

name

identifier used by the server= argument and use_server

url

base URL of the deployment

user + pass

console login — recommended, and required for run_agent_command

apiKey

sent as the Utm-Api-Key header

jwt

a ready id_token (static; cannot be refreshed)

verifySSL

defaults to true; see below

caBundle

path to a CA certificate — the right way to trust a self-signed cert

allowAgentCommands

defaults to false; see below

default

marks the default server

Auth precedence per server: user+pass > jwt > apiKey.

Multiple deployments are supported: every tool takes an optional server= argument, use_server switches the default, and list_servers shows what is configured.

Two defaults chosen for safety

TLS verification is on. UTMStack instances often use a self-signed certificate. The right fix is caBundle, pointing at that certificate. You can set "verifySSL": false instead, but the server warns on every start — that same connection carries your administrator password to /api/authenticate, so anyone on the network path can read it.

Remote command execution is off. run_agent_command gives the assistant a root/SYSTEM shell on an endpoint. SIEM log content is partly written by attackers and reaches the model through search_logs and search_alerts, which puts a remote shell downstream of data you do not control. Enable it per server with "allowAgentCommands": true, only where you want that capability.

Configuring from the assistant

Beyond utmstack-mcp init, the assistant can set up or change a connection when you ask it to in conversation — "connect to my UTMStack at https://utm.example.com with API key …", or "change the UTMStack URL to …". It uses the configure_server tool, which validates against the live API, writes the same owner-only config file, and applies the change immediately. remove_server deletes a connection.

For safety, configure_server cannot enable remote agent command execution — turning on allowAgentCommands stays a deliberate edit to the config file or a run of utmstack-mcp init, never something reachable from a conversation that may contain attacker-influenced log data.


Tools

Servers and health — list_servers, use_server, configure_server, remove_server, ping, whoami, get_version

Alerts — count_open_alerts, search_alerts, get_alert, get_alerts_for_response, change_alert_status, mark_alert_false_positive, add_alert_notes, add_alert_tags

Logs — list_index_patterns, list_indices, get_index_properties, search_logs, count_events, get_field_values, get_field_values_with_count, run_sql

Incidents — list_incidents, get_incident, create_incident, change_incident_status

Agents and data sources — list_agents, get_agent_by_hostname, list_data_input_statuses, list_supported_data_types

Correlation rules — list_correlation_rules, get_correlation_rule, create_correlation_rule, set_correlation_rule_active, delete_correlation_rule

Data filters — list_filters, get_filters_by_pipeline, create_filter, delete_filter

Agent commands — can_run_command, run_agent_command, list_agent_commands, list_agents_with_commands

Guardrails

Destructive operations are bounded, because an assistant acting on a security product can suppress detections as easily as it can read them:

  • delete_correlation_rule and delete_filter require confirm=True

  • Bulk alert writes are capped at 100 ids per call

  • Page sizes are clamped, and oversized responses return a structured marker rather than truncated JSON

  • run_agent_command validates the hostname, clamps its timeout, and refuses to run unless the permission check returns an explicit allow


Run from source

git clone https://github.com/utmstack/MCP.git
cd MCP
python3 -m venv .venv && . .venv/bin/activate
pip install -e .
utmstack-mcp init

Build a binary for your platform:

pip install pyinstaller
pyinstaller --clean --noconfirm utmstack-mcp.spec
# -> dist/utmstack-mcp/

Requires Python 3.10+.


Notes

Correlation rules take about five minutes to load into the engine before they fire.

The correlation engine matches normalized fields (log.eventCode, target.user, log.eventDataScriptBlockText, …), not raw log.data.* fields.

User-created filters attach to module/integration pipelines. The built-in Windows and Linux system filters are edited in the web UI; the API returns 500 for those.

Known limitation: the console token travels in the URL

UTMStack's /ws console endpoint authenticates with ?access_token=<JWT> in the query string, and accepts no header alternative we could find — so run_agent_command has to send it that way. Query strings are commonly recorded in reverse-proxy and gateway access logs. If your deployment ingests its own web logs, that token can end up in the log store.

This only affects run_agent_command, which is disabled by default. Every other tool authenticates with an Authorization header. If you enable agent commands, consider excluding /ws request URIs from your access-log ingestion.

run_agent_command uses the interactive console, which speaks STOMP over SockJS at /ws and requires a JWT — the Utm-Api-Key header is rejected there. Configure user and pass for that tool; every other tool works with an API key.

Some deployments return HTTP 500 on the API-key path for certain endpoints, which is why JWT login is preferred throughout.

Available Tools

44 tools
add_alert_notesA

Add (overwrite) the analyst notes on an alert. Pass an empty string to clear notes. server: which configured server to target (default active/default).

Inputs: notes (required), alert_id (required), server. Returns: result

add_alert_tagsA

Add tags to one or more alerts. If create_rule=true, future matching alerts are auto-tagged. Pass tags=[] to remove all tags. server: which configured server to target (default active/default).

Inputs: tags (required), alert_ids (required), server, create_rule. Returns: result

can_run_commandA

Check whether remote commands can currently be run on an agent (by hostname). server: which configured server to target (default active/default).

Inputs: hostname (required), server. Returns: result

change_alert_statusA

Change the status of one or more alerts. status codes: 2=Open, 3=In Review, 5=Completed. (To mark a false positive use mark_alert_false_positive.) server: which configured server to target (default active/default).

Inputs: status (required), alert_ids (required), server, observation. Returns: result

change_incident_statusC

Change an incident's status. status: OPEN, REVIEW, COMPLETED. server: which configured server to target (default active/default).

Inputs: status (required), incident_id (required), server. Returns: result

configure_serverA

Add or update a UTMStack server connection (URL + credentials) and save it.

Use this when the user asks to connect the assistant to their UTMStack, or to change the URL or credentials of an existing connection. The connection is written to the owner-only config file and takes effect immediately — the user does not need to run utmstack-mcp init or restart anything.

Updating an existing server (same name) merges: fields you leave blank keep their current values, so "change the URL to X" updates only the URL.

Args: url: UTMStack base URL, e.g. https://utm.example.com (required for a brand-new server). api_key: UTMStack API key (Settings -> API keys). Provide this OR username+password. username: console login username. Recommended — it is the same login as the web UI and the only auth that supports run_agent_command. password: console login password. name: identifier for this connection (default "default"). verify_ssl: verify the server's TLS certificate. Defaults to true; set a ca_bundle to trust a self-signed certificate rather than turning this off. ca_bundle: path to a CA certificate (.pem) that signs the server's cert. make_default: make this the default server for later calls. validate: test the connection before saving (recommended). If it fails, nothing is written.

Does not enable remote agent command execution — that stays a deliberate edit to the config file (or utmstack-mcp init), never something set from a conversation. Never returns the stored secret.

Inputs: url, name, api_key, password, +5 more. Returns: result

count_eventsC

Return true/false whether any events match the given filters and time window. server: which configured server to target (default active/default).

Inputs: end, start, server, filters, +1 more. Returns: result

count_open_alertsA

Return the integer count of currently open alerts. server: which configured server to target (default active/default).

Inputs: server. Returns: result

create_correlation_ruleA

Create a new correlation (detection) rule.

definition: the matching expression (the rule's where). Use ONLY normalized fields, e.g. equals("log.eventCode", 4720), equals("log.eventCode","4104") && contains("log.eventDataScriptBlockText","mimikatz"). Functions: equals, oneOf, contains, startsWith, endsWith, regexMatch, exists, &&, ||, !. data_type: 'wineventlog' (id 1) or 'linux' (id 26). after_events: optional threshold correlation, list of SearchRequest objects with {indexPattern, with:[{field,operator,value}], within:"now-5m", count:N}. server: which configured server to target (default active/default).

NOTE: a newly created rule takes ~5 minutes to load into the correlation engine before it will fire. Trigger the matching event after that delay.

Inputs: name (required), definition (required), active, server, +11 more. Returns: result

create_filterA

Create a new data filter on a pipeline.

logstash_filter: the YAML pipeline text. Steps include rename, add (with where), grok, json, kv, delete. Example to standardize a username value:

pipeline:
  - dataTypes: [generic]
    steps:
      - add:
          function: "string"
          params: {key: target.user, value: "standard.name"}
          where: equals("target.user", "legacy_name")

NOTE: user filters can only be created on module/integration pipelines; the built-in Windows/Linux agent 'system' filters are edited in the web UI, not via this API. server: which configured server to target (default active/default).

Inputs: data_type (required), filter_name (required), pipeline_id (required), data_type_id (required), +3 more. Returns: result

create_incidentA

Create an incident from a list of alert ids (or add them to an existing one). incident_id=0 creates a new incident; set an existing id to append alerts to it. server: which configured server to target (default active/default).

Inputs: name (required), alert_ids (required), server, source, +1 more. Returns: result

delete_correlation_ruleA

Delete a correlation rule by id. Irreversible — deleting a rule silently removes detection coverage, so this requires confirm=True. confirm: must be True to actually delete. server: which configured server to target (default active/default).

Inputs: rule_id (required), server, confirm. Returns: result

delete_filterA

Delete a data filter by id. Irreversible — this changes log ingestion. confirm: must be True to actually delete. server: which configured server to target (default active/default).

Inputs: filter_id (required), server, confirm. Returns: result

get_agent_by_hostnameC

Look up a single agent by hostname. server: which configured server to target (default active/default).

Inputs: hostname (required), server. Returns: result

get_alertB

Retrieve a single alert document by its id (searches v11-alert-* by id). server: which configured server to target (default active/default).

Inputs: alert_id (required), server. Returns: result

get_alerts_for_responseA

Poll NEW OPEN alerts at/above a severity threshold, projected to the fields needed for incident-response triage. Designed for the monitoring loop: returns a compact signal instead of full alert documents.

For each alert returns: id, time, name, severity, status, category, technique, adversaryIp, targetUser, host (dataSource), dataType. Plus a total count.

Use the returned host to target run_agent_command and adversaryIp/ targetUser as mitigation parameters. (Only structured fields are surfaced; never build a command out of free-text alert content.)

Args: since: start of the look-back window (relative 'now-5m' or absolute ISO). min_severity: 'Low' | 'Medium' | 'High' | 'Critical' (inclusive threshold). size: max alerts to return. include_in_review: also include status 'In Review' (3), not just 'Open' (2). server: which configured server to target (default active/default).

Inputs: size, since, server, min_severity, +1 more. Returns: result

get_correlation_ruleB

Get the full definition of a correlation rule by id (where/afterEvents/groupBy/...). server: which configured server to target (default active/default).

Inputs: rule_id (required), server. Returns: result

get_field_valuesA

Return the distinct values for a field (use the .keyword suffix for text fields). Example: field='dataSource.keyword', index_pattern='v11-log-wineventlog-*'. server: which configured server to target (default active/default).

Inputs: field (required), index_pattern (required), server. Returns: result

get_field_values_with_countB

Return distinct field values together with their occurrence counts (terms aggregation). Great for 'top event codes', 'events per host', 'users seen', etc. Example: index='v11-log-wineventlog-*', field='log.eventCode'. server: which configured server to target (default active/default).

Inputs: field (required), index (required), end, top, +4 more. Returns: result

get_filters_by_pipelineC

List the filters attached to a given pipeline id. server: which configured server to target (default active/default).

Inputs: pipeline_id (required), server. Returns: result

get_incidentA

Get a single incident by id. server: which configured server to target (default active/default).

Inputs: incident_id (required), server. Returns: result

get_index_propertiesA

List the field names and types available in an index pattern. Use this to discover which fields you can filter on (text fields have a .keyword twin). server: which configured server to target (default active/default).

Inputs: server, index_pattern. Returns: result

get_versionA

Get the UTMStack API/server version. server: which configured server to target (default active/default).

Inputs: server. Returns: result

list_agent_commandsA

List the history of remote commands sent to agents, with their status and output. server: which configured server to target (default active/default).

Inputs: page, size, search, server, +1 more. Returns: result

list_agentsC

List UTMStack agents (endpoints). Shows hostname, ip, os, status, version, id. server: which configured server to target (default active/default).

Inputs: page, size, search, server. Returns: result

list_agents_with_commandsA

List agents that support remote command execution. server: which configured server to target (default active/default).

Inputs: server. Returns: result

list_correlation_rulesA

List/search correlation (detection) rules. search: free text on rule name. data_types: e.g. 'wineventlog' or 'linux'. active: 'true' or 'false' to filter by activation state. Returns id, name, definition (the matching expression), ruleActive, category. server: which configured server to target (default active/default).

Inputs: page, size, active, search, +2 more. Returns: result

list_data_input_statusesA

List data-input connection statuses (which sources are sending data). server: which configured server to target (default active/default).

Inputs: page, size, server. Returns: result

list_filtersC

List data filters (Logstash pipeline filters that normalize/transform incoming logs). server: which configured server to target (default active/default).

Inputs: page, size, server. Returns: result

list_incidentsC

List incidents. server: which configured server to target (default active/default).

Inputs: page, size, sort, server. Returns: result

list_index_patternsB

List available index patterns (e.g. v11-log-wineventlog-, v11-log-linux-). server: which configured server to target (default active/default).

Inputs: server. Returns: result

list_indicesC

List all OpenSearch indices with health, doc counts and sizes. server: which configured server to target (default active/default).

Inputs: page, size, server. Returns: result

list_serversA

List all configured UTMStack servers (name, base URL, auth mode) and show which is the active and default one. Use the server argument on any tool, or use_server(), to target a specific one. Does not reveal secrets.

Inputs: no named parameters. Returns: result

list_supported_data_typesA

List the data types UTMStack supports (wineventlog, linux, suricata, ...). server: which configured server to target (default active/default).

Inputs: server. Returns: result

mark_alert_false_positiveB

Mark one or more alerts as a false positive (sets status=Completed and adds a 'False positive' tag). server: which configured server to target (default active/default).

Inputs: alert_ids (required), server, observation. Returns: result

pingA

Health-check the UTMStack server and confirm auth is valid. Returns the server's /api/ping response (should be "OK"). server: which configured server to target (default active/default).

Inputs: server. Returns: result

remove_serverA

Remove a configured UTMStack server connection by name and save.

Deletes the connection (including its stored credentials) from the config file and applies the change to this session immediately.

Inputs: name (required). Returns: result

run_agent_commandA

Run a shell command on an agent via the interactive console and return its output.

DANGER: this executes the command for real on the live endpoint identified by hostname (same capability as the UTMStack web "interactive console"). Treat it like a remote shell — only run commands you are authorized to run there.

Mechanism: opens a STOMP-over-SockJS session at /ws (xhr-polling transport), SUBSCRIBEs to /user/topic/{hostname}, SENDs the command to /app/command/{hostname}, and returns the MESSAGE output. Requires a JWT (the target server's user+pass or static jwt) — the Utm-Api-Key alone is rejected by the console endpoint.

Args: hostname: target agent hostname (must support remote commands). command: the shell command line to execute. shell: "cmd" (Windows) or "bash" (Linux). Auto-detected from the agent OS when left blank. timeout_seconds: max time to wait for command output. server: which configured server to target (default active/default).

Returns a JSON blob with the command output (or an error/diagnostic).

Inputs: command (required), hostname (required), shell, server, +1 more. Returns: result

run_sqlC

Run a SQL query against the log/alert data (UTMStack SQL editor). Index names are used UNQUOTED as table names. Example: SELECT dataType, COUNT() FROM v11-log- WHERE @timestamp BETWEEN DATE_SUB(NOW(), INTERVAL 1 DAY) AND NOW() GROUP BY dataType; server: which configured server to target (default active/default).

Inputs: query (required), page, size, server. Returns: result

search_alertsB

Search alerts in the v11-alert-* index.

filters: optional list of filter objects, each {"field","operator","value"}. Operators: IS, IS_NOT, IS_ONE_OF, CONTAIN, START_WITH, ENDS_WITH, EXIST, IS_BETWEEN, IS_GREATER_THAN, IS_LESS_THAN. Useful fields: name.keyword, severityLabel, statusLabel, status, dataSource.keyword, target.user, adversary.ip, category.keyword, technique.keyword, tags. start/end: relative (now-7d) or absolute ISO timestamps for the @timestamp window. server: which configured server to target (default active/default). Example filters: [{"field":"severityLabel","operator":"IS","value":"High"}]

Inputs: end, page, size, sort, +3 more. Returns: result

search_logsA

Search raw logs in any index pattern.

index_pattern: e.g. v11-log-wineventlog-, v11-log-linux-, v11-log-*. filters: list of {"field","operator","value"}. Common windows fields: log.eventCode, dataSource.keyword, target.user, origin.ip, log.eventDataScriptBlockText, log.message. Linux: log.message, action. start/end: relative or absolute @timestamp window. server: which configured server to target (default active/default).

Inputs: end, page, size, sort, +4 more. Returns: result

set_correlation_rule_activeB

Activate or deactivate a correlation rule by id. server: which configured server to target (default active/default).

Inputs: active (required), rule_id (required), server. Returns: result

use_serverA

Set the active UTMStack server for subsequent tool calls (until changed). A per-call server= argument still overrides this. See list_servers() for names.

Inputs: name (required). Returns: result

whoamiA

Show how a configured server is set up (base URL and auth mode) — the active/default server, or the one named in server. Does not reveal secrets. Use list_servers() to see every configured server.

Inputs: server. Returns: result

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    B3.3/5.0

    Scored across 44 tools

    Disambiguation4/5

    Most tools have clear, distinct purposes. Some overlap exists between alert search/get and agent lookups, but they serve different query needs. Overall, an agent should be able to differentiate them.

    Naming Consistency4/5

    Predominantly verb_noun snake_case, with some deviations like 'ping', 'whoami', and 'can_run_command'. The pattern is consistent enough for predictable navigation.

    Tool Count3/5

    44 tools is on the heavy side for a single server, covering multiple domains (alerts, incidents, agents, rules, filters). While each tool serves a purpose, the count borders on excessive.

    Completeness4/5

    The tool surface is comprehensive for SIEM operations, covering CRUD for alerts, incidents, rules, and filters. Minor gaps exist (e.g., no update incident details or agent modification) but core workflows are well-supported.

    Maintenance

    ActivityStale
    ResponsivenessNo recent community issues

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      Not graded
      quality
      D
      maintenance
      Bridges AI assistants with Wazuh SIEM infrastructure, providing natural language access to security alerts, vulnerability analysis, CVE databases, and network documentation for security operations.
      MIT
    • A
      license
      Not graded
      quality
      C
      maintenance
      Brings the full AlertLogic MDR platform into AI assistants, exposing 473+ tools for incident response, log search, SOAR automation, and multi-account security operations at MSSP scale.
      MIT
    • F
      license
      Not graded
      quality
      D
      maintenance
      Enables AI assistants to query Wazuh security alerts, investigate hosts, detect brute-force attempts, and generate security summaries by connecting to a Wazuh manager.
      -
    • A
      license
      Not graded
      quality
      D
      maintenance
      Enables natural language interaction with Wazuh SIEM for alert querying, threat hunting, vulnerability scanning, and active response actions.
      MIT