Provides AI assistants with structured access to the LogicMonitor REST API to manage alerts, devices, dashboards, and network metrics.
Also known as ryanmat/mcp-server-logicmonitor, mcp-server-logicmonitor, lm-mcp
- Stars
- 2
- Tools
- 1
- Downloads, 30 days
- 846
- Latest version
- 5.0.1
- Last commit
- Nov 14, 2026
- License
- MIT
What is LogicMonitor MCP?
LogicMonitor provides AI assistants with structured access to the LogicMonitor REST API to manage alerts, devices, dashboards, and network metrics. It exposes 1 tool, such as search_tools.
Summary generated from the project's README
About LogicMonitor MCP server
Managing network infrastructure and monitoring alerts through natural language is what this server enables for DevOps and IT operations teams. It provides AI assistants with structured access to the LogicMonitor REST API v3, offering 244 tools to interact with devices, dashboards, metrics, and collectors. The server runs as a local subprocess via stdio, making it compatible with clients like Claude Desktop, Cursor, and Claude Code. Users must provide a LogicMonitor portal URL and a bearer token via environment variables to establish a connection.
By default, the server operates in a read-only mode to protect your environment. To perform actions like acknowledging alerts, creating dashboards, or managing scheduled downtime, you must explicitly enable write operations. The server includes specialized composite tools such as triage, diagnose, and capacity_plan that combine multiple API calls into single responses. It also supports advanced analysis like metric forecasting, topology-based blast radius assessment, and Terraform HCL generation for infrastructure as code workflows.
What you can do with it
- 1Automated alert investigationAsk your assistant to investigate active alerts, score noise, and assess the blast radius of an incident.
- 2Predictive capacity planningRequest a forecast to determine when specific device metrics like memory will reach a critical threshold.
- 3Infrastructure as Code generationConvert existing LogicMonitor portal resources into HCL code for use with Terraform.
Written from the project's README. Every use case is checked against a quote from it.
How to install LogicMonitor MCP
Formats verified 10 Nov 2026Add to CursorAdd to GooseInstall LogicMonitor MCP in Claude Code
$ claude mcp add --env 'LM_BEARER_TOKEN=<YOUR_LM_BEARER_TOKEN>' --env 'LM_PORTAL=<LM_PORTAL>' --env 'EDWIN_CLIENT_TOKEN=<YOUR_EDWIN_CLIENT_TOKEN>' --env 'LM_ACCESS_KEY=<YOUR_LM_ACCESS_KEY>' --env 'LM_HTTP_AUTH_TOKEN=<YOUR_LM_HTTP_AUTH_TOKEN>' --env 'LM_HTTP_SSL_KEYFILE_PASSWORD=<YOUR_LM_HTTP_SSL_KEYFILE_PASSWORD>' mcp-server-logicmonitor -- uvx lm-mcpInstall LogicMonitor MCP in Cursor
~/.cursor/mcp.json
{
"mcpServers": {
"mcp-server-logicmonitor": {
"command": "uvx",
"args": [
"lm-mcp"
],
"env": {
"LM_BEARER_TOKEN": "<YOUR_LM_BEARER_TOKEN>",
"LM_PORTAL": "<LM_PORTAL>",
"EDWIN_CLIENT_TOKEN": "<YOUR_EDWIN_CLIENT_TOKEN>",
"LM_ACCESS_KEY": "<YOUR_LM_ACCESS_KEY>",
"LM_HTTP_AUTH_TOKEN": "<YOUR_LM_HTTP_AUTH_TOKEN>",
"LM_HTTP_SSL_KEYFILE_PASSWORD": "<YOUR_LM_HTTP_SSL_KEYFILE_PASSWORD>"
}
}
}
}Install LogicMonitor MCP in VS Code
.vscode/mcp.json
{
"servers": {
"mcp-server-logicmonitor": {
"type": "stdio",
"command": "uvx",
"args": [
"lm-mcp"
],
"env": {
"LM_BEARER_TOKEN": "<YOUR_LM_BEARER_TOKEN>",
"LM_PORTAL": "<LM_PORTAL>",
"EDWIN_CLIENT_TOKEN": "<YOUR_EDWIN_CLIENT_TOKEN>",
"LM_ACCESS_KEY": "<YOUR_LM_ACCESS_KEY>",
"LM_HTTP_AUTH_TOKEN": "<YOUR_LM_HTTP_AUTH_TOKEN>",
"LM_HTTP_SSL_KEYFILE_PASSWORD": "<YOUR_LM_HTTP_SSL_KEYFILE_PASSWORD>"
}
}
}
}Install LogicMonitor MCP in VS Code CLI
$ code --add-mcp '{"name":"mcp-server-logicmonitor","type":"stdio","command":"uvx","args":["lm-mcp"],"env":{"LM_BEARER_TOKEN":"<YOUR_LM_BEARER_TOKEN>","LM_PORTAL":"<LM_PORTAL>","EDWIN_CLIENT_TOKEN":"<YOUR_EDWIN_CLIENT_TOKEN>","LM_ACCESS_KEY":"<YOUR_LM_ACCESS_KEY>","LM_HTTP_AUTH_TOKEN":"<YOUR_LM_HTTP_AUTH_TOKEN>","LM_HTTP_SSL_KEYFILE_PASSWORD":"<YOUR_LM_HTTP_SSL_KEYFILE_PASSWORD>"}}'Install LogicMonitor MCP in Claude Desktop
claude_desktop_config.json
{
"mcpServers": {
"mcp-server-logicmonitor": {
"command": "uvx",
"args": [
"lm-mcp"
],
"env": {
"LM_BEARER_TOKEN": "<YOUR_LM_BEARER_TOKEN>",
"LM_PORTAL": "<LM_PORTAL>",
"EDWIN_CLIENT_TOKEN": "<YOUR_EDWIN_CLIENT_TOKEN>",
"LM_ACCESS_KEY": "<YOUR_LM_ACCESS_KEY>",
"LM_HTTP_AUTH_TOKEN": "<YOUR_LM_HTTP_AUTH_TOKEN>",
"LM_HTTP_SSL_KEYFILE_PASSWORD": "<YOUR_LM_HTTP_SSL_KEYFILE_PASSWORD>"
}
}
}
}Install LogicMonitor MCP in Codex
$ codex mcp add mcp-server-logicmonitor --env 'LM_BEARER_TOKEN=<YOUR_LM_BEARER_TOKEN>' --env 'LM_PORTAL=<LM_PORTAL>' --env 'EDWIN_CLIENT_TOKEN=<YOUR_EDWIN_CLIENT_TOKEN>' --env 'LM_ACCESS_KEY=<YOUR_LM_ACCESS_KEY>' --env 'LM_HTTP_AUTH_TOKEN=<YOUR_LM_HTTP_AUTH_TOKEN>' --env 'LM_HTTP_SSL_KEYFILE_PASSWORD=<YOUR_LM_HTTP_SSL_KEYFILE_PASSWORD>' -- uvx lm-mcpInstall LogicMonitor MCP in Codex config.toml
~/.codex/config.toml
[mcp_servers.mcp-server-logicmonitor] command = "uvx" args = ["lm-mcp"] [mcp_servers.mcp-server-logicmonitor.env] LM_BEARER_TOKEN = "<YOUR_LM_BEARER_TOKEN>" LM_PORTAL = "<LM_PORTAL>" EDWIN_CLIENT_TOKEN = "<YOUR_EDWIN_CLIENT_TOKEN>" LM_ACCESS_KEY = "<YOUR_LM_ACCESS_KEY>" LM_HTTP_AUTH_TOKEN = "<YOUR_LM_HTTP_AUTH_TOKEN>" LM_HTTP_SSL_KEYFILE_PASSWORD = "<YOUR_LM_HTTP_SSL_KEYFILE_PASSWORD>"
Install LogicMonitor MCP in Gemini CLI
$ gemini mcp add -s user -e 'LM_BEARER_TOKEN=<YOUR_LM_BEARER_TOKEN>' -e 'LM_PORTAL=<LM_PORTAL>' -e 'EDWIN_CLIENT_TOKEN=<YOUR_EDWIN_CLIENT_TOKEN>' -e 'LM_ACCESS_KEY=<YOUR_LM_ACCESS_KEY>' -e 'LM_HTTP_AUTH_TOKEN=<YOUR_LM_HTTP_AUTH_TOKEN>' -e 'LM_HTTP_SSL_KEYFILE_PASSWORD=<YOUR_LM_HTTP_SSL_KEYFILE_PASSWORD>' mcp-server-logicmonitor uvx -- lm-mcpInstall LogicMonitor MCP in Gemini settings.json
~/.gemini/settings.json
{
"mcpServers": {
"mcp-server-logicmonitor": {
"command": "uvx",
"args": [
"lm-mcp"
],
"env": {
"LM_BEARER_TOKEN": "<YOUR_LM_BEARER_TOKEN>",
"LM_PORTAL": "<LM_PORTAL>",
"EDWIN_CLIENT_TOKEN": "<YOUR_EDWIN_CLIENT_TOKEN>",
"LM_ACCESS_KEY": "<YOUR_LM_ACCESS_KEY>",
"LM_HTTP_AUTH_TOKEN": "<YOUR_LM_HTTP_AUTH_TOKEN>",
"LM_HTTP_SSL_KEYFILE_PASSWORD": "<YOUR_LM_HTTP_SSL_KEYFILE_PASSWORD>"
}
}
}
}Needs a key before it starts
Started on 16 Nov 2026 with no keys, it stopped and asked for one, so a client sees no tools until you set the variables above.
LogicMonitor MCP tools 1
Detected from the READMEsearch_tools
| search_tools | — |
Configuration
LM_BEARER_TOKENRequiredSecretLogicMonitor API Bearer token; LM_ACCESS_ID with LM_ACCESS_KEY can stand in for it
LM_PORTALRequiredPortal hostname, such as company.logicmonitor.com; required unless LM_MULTI_PORTAL is true
EDWIN_CLIENT_TOKENSecretSecret of that Edwin AI credential
LM_ACCESS_KEYSecretLMv1 API access key, the secret of the LM_ACCESS_ID token
LM_HTTP_AUTH_TOKENSecretBearer token every HTTP request must carry, except to the health endpoints and /; 16 characters or more
LM_HTTP_SSL_KEYFILE_PASSWORDSecretPassword of an encrypted LM_HTTP_SSL_KEYFILE
EDWIN_CLIENT_IDEdwin AI API credential client id; with EDWIN_CLIENT_TOKEN it turns on the eight Edwin tools
EDWIN_HOSTEdwin AI base URL, https only (default: https://<portal name>.dexda.ai; required in multi-portal mode)
LM_ACCESS_IDLMv1 API access ID; with LM_ACCESS_KEY it stands in for LM_BEARER_TOKEN
LM_AGE_KEYAge identity file that decrypts LM_VAULT_FILE
LM_DISABLED_TOOLSComma-separated tool names or glob patterns to disable (e.g., delete_*)
LM_ENABLED_TOOLSComma-separated tool names or glob patterns to enable (e.g., get_*,create_device)
LM_ENABLE_WRITE_OPERATIONSList and allow the tools that change the portal (create, update, delete, acknowledge, run)
LM_HEALTH_CHECK_CONNECTIVITYAdd a LogicMonitor API call to the HTTP /health check for callers that pass the bearer check
LM_HTTP_ALLOWED_HOSTSComma-separated Host names the HTTP server accepts besides the loopback names and LM_HTTP_HOST
LM_HTTP_ALLOW_UNAUTHENTICATEDAccept a non-loopback HTTP bind with no LM_HTTP_AUTH_TOKEN, for a listener another layer protects
LM_HTTP_HOSTHTTP bind address; a non-loopback bind needs LM_HTTP_AUTH_TOKEN or LM_HTTP_ALLOW_UNAUTHENTICATED
LM_HTTP_PORTHTTP port
LM_HTTP_SSL_CERTFILETLS certificate for the HTTP transport
LM_HTTP_SSL_KEYFILETLS private key for the HTTP transport
LM_LOG_LEVELServer log level
LM_MAX_RETRIESResends of a rate-limited request, or of a 5xx on an idempotent method, 0 to 10
LM_MCP_CATEGORIESComma-separated categories to list (read, write, delete, export, import, session, workflow); narrows the tool filters, never widens them
LM_MULTI_PORTALServe several portals from one server, chosen at runtime (stdio only)
LM_PORTALS_FILEPlaintext JSON portal map for testing multi-portal mode; LM_VAULT_FILE wins when both are set
LM_RETRY_AFTER_CAP_SECONDSLongest wait in seconds a server's Retry-After header can impose between retries
LM_SESSION_HISTORY_SIZETool calls kept in the session history, 0 to 1000
LM_TIMEOUTAPI request timeout in seconds, 5 to 300
LM_TRANSPORTstdio, or http for the Streamable HTTP endpoint at /mcp
LM_VAULT_FILEAge-encrypted portal vault (multi-portal mode)
| Name | Description | |
|---|---|---|
| LM_BEARER_TOKEN | LogicMonitor API Bearer token; LM_ACCESS_ID with LM_ACCESS_KEY can stand in for it | RequiredSecret |
| LM_PORTAL | Portal hostname, such as company.logicmonitor.com; required unless LM_MULTI_PORTAL is true | Required |
| EDWIN_CLIENT_TOKEN | Secret of that Edwin AI credential | Secret |
| LM_ACCESS_KEY | LMv1 API access key, the secret of the LM_ACCESS_ID token | Secret |
| LM_HTTP_AUTH_TOKEN | Bearer token every HTTP request must carry, except to the health endpoints and /; 16 characters or more | Secret |
| LM_HTTP_SSL_KEYFILE_PASSWORD | Password of an encrypted LM_HTTP_SSL_KEYFILE | Secret |
| EDWIN_CLIENT_ID | Edwin AI API credential client id; with EDWIN_CLIENT_TOKEN it turns on the eight Edwin tools | |
| EDWIN_HOST | Edwin AI base URL, https only (default: https://<portal name>.dexda.ai; required in multi-portal mode) | |
| LM_ACCESS_ID | LMv1 API access ID; with LM_ACCESS_KEY it stands in for LM_BEARER_TOKEN | |
| LM_AGE_KEY | Age identity file that decrypts LM_VAULT_FILE | |
| LM_DISABLED_TOOLS | Comma-separated tool names or glob patterns to disable (e.g., delete_*) | |
| LM_ENABLED_TOOLS | Comma-separated tool names or glob patterns to enable (e.g., get_*,create_device) | |
| LM_ENABLE_WRITE_OPERATIONS | List and allow the tools that change the portal (create, update, delete, acknowledge, run) | |
| LM_HEALTH_CHECK_CONNECTIVITY | Add a LogicMonitor API call to the HTTP /health check for callers that pass the bearer check | |
| LM_HTTP_ALLOWED_HOSTS | Comma-separated Host names the HTTP server accepts besides the loopback names and LM_HTTP_HOST | |
| LM_HTTP_ALLOW_UNAUTHENTICATED | Accept a non-loopback HTTP bind with no LM_HTTP_AUTH_TOKEN, for a listener another layer protects | |
| LM_HTTP_HOST | HTTP bind address; a non-loopback bind needs LM_HTTP_AUTH_TOKEN or LM_HTTP_ALLOW_UNAUTHENTICATED | |
| LM_HTTP_PORT | HTTP port | |
| LM_HTTP_SSL_CERTFILE | TLS certificate for the HTTP transport | |
| LM_HTTP_SSL_KEYFILE | TLS private key for the HTTP transport | |
| LM_LOG_LEVEL | Server log level | |
| LM_MAX_RETRIES | Resends of a rate-limited request, or of a 5xx on an idempotent method, 0 to 10 | |
| LM_MCP_CATEGORIES | Comma-separated categories to list (read, write, delete, export, import, session, workflow); narrows the tool filters, never widens them | |
| LM_MULTI_PORTAL | Serve several portals from one server, chosen at runtime (stdio only) | |
| LM_PORTALS_FILE | Plaintext JSON portal map for testing multi-portal mode; LM_VAULT_FILE wins when both are set | |
| LM_RETRY_AFTER_CAP_SECONDS | Longest wait in seconds a server's Retry-After header can impose between retries | |
| LM_SESSION_HISTORY_SIZE | Tool calls kept in the session history, 0 to 1000 | |
| LM_TIMEOUT | API request timeout in seconds, 5 to 300 | |
| LM_TRANSPORT | stdio, or http for the Streamable HTTP endpoint at /mcp | |
| LM_VAULT_FILE | Age-encrypted portal vault (multi-portal mode) |
Automated checks
Ran Nov 12, 2026Full report →Passed our automated checks. That is not a security audit: read the code and give the server the narrowest access it needs. How checks work
- Source repository linked and reachablePass
- Open source license (OSI)Pass
- Commit or release in the last yearPass
- Package name not close to a popular packagePass
- Remote endpoint uses HTTPSNot checked
- Destructive tools are labelledPass
- No hidden instructions to the modelPass
- No critical advisory in the package or its dependenciesPass
README
View on GitHubLogicMonitor MCP Server
Model Context Protocol (MCP) server that gives AI assistants structured access to the LogicMonitor REST API v3 through 258 tools, 3 workflow prompts, and 25 resources.
Overview
The server runs as a local subprocess of your MCP client and calls your portal with your API token. It works with any MCP client that launches a stdio server, including Claude Desktop, Claude Code, Cursor, Continue, and Cline.
Until you set LM_ENABLE_WRITE_OPERATIONS=true, the server does not list the tools that change your portal. Composite tools such as triage, diagnose, health_check, portal_overview, capacity_plan, and detect_site_outage answer common investigations in one call, and search_tools finds the right tool when your client cannot load them all.
The 258 counts every tool the server can advertise; four of them appear only in multi-portal mode and eight only with Edwin AI credentials; see Integrations.
What's new in v4.3.0
A hardening release: mostly fixes to HTTP exposure, Terraform gating, and analytics that returned wrong or partial answers without saying so.
Check these before you upgrade:
LM_TRANSPORT=httpwith noLM_HTTP_AUTH_TOKENrefuses to start on the default bind, and HTTP routes checkHostandContent-Type.terraform_initandterraform_planneed writes enabled, theterraformsubprocess no longer inherits cloud credentials, andbackend_configis an object.get_topology_mapno longer takesdevice_group_id.get_collector_healthreads each collector'sisDownfield and no longer returnsactive_collector_down_alerts, sodetect_site_outageverdicts change where a collector is down.calculate_availabilityandscore_alert_noiseread cleared alerts, so availability falls and flap counts rise where cleared incidents exist.
The CHANGELOG has every change and what to do about it.
- v4.2.1: LMv1 portals authenticate again, tool errors reach the client with
isError: true, every tool is callable per its own schema, and the container installs fromuv.lock. - v4.2.0: Multi-portal mode serves many customer portals from one server, with credentials in an age-encrypted vault.
- v4.1.0: Native OTLP metrics tools, for discovery and PromQL range queries.
- v4.0.0: Tools that called nonexistent API paths are gone, the cost and Service Insight tools move to working APIs, and the Automated Diagnostics and Remediation surface arrives.
Setup
You need uv for uvx, and a LogicMonitor API token.
To connect Claude Code:
-
In your portal, go to Settings > Users and Roles > API Tokens.
-
Create an API-only user or add a token to an existing user, then copy the bearer token.
-
Add the server:
claude mcp add logicmonitor \ -e LM_PORTAL=yourcompany.logicmonitor.com \ -e LM_BEARER_TOKEN=<your-bearer-token> \ -- uvx --from lm-mcp lm-mcp-server -
Run
claude mcp list. The output listslogicmonitoras connected. -
Ask your assistant to "List the first 5 devices in LogicMonitor".
Cursor, Claude Desktop, Cline, and Gemini CLI run the same server with the same block:
{
"mcpServers": {
"logicmonitor": {
"command": "uvx",
"args": ["--from", "lm-mcp", "lm-mcp-server"],
"env": {
"LM_PORTAL": "yourcompany.logicmonitor.com",
"LM_BEARER_TOKEN": "your-bearer-token"
}
}
}
}
documentation/client-setup.md has each client's config file location and exact steps, including OpenAI Codex and GitHub Copilot, whose formats differ. Cursor loads only the first 40 MCP tools, and that page shows how to fit a curated set under the cap with LM_MCP_CATEGORIES.
Other install methods
# pip
pip install lm-mcp
# From source
git clone https://github.com/ryanmat/mcp-server-logicmonitor.git
cd mcp-server-logicmonitor
uv sync
A uv tool install does not update itself: run uv tool upgrade lm-mcp, then restart your MCP client.
Usage
Ask your assistant in natural language:
- "Show me all critical alerts from the last hour"
- "Acknowledge alert LMA12345 with note 'Investigating disk issue'"
- "What datasources are applied to device 123, and get CPU metrics for the last hour"
- "Create a dashboard called 'API Health' and add a graph widget"
- "Create a 2-hour maintenance window (SDT) for device 123"
- "Triage all critical alerts from the last 4 hours"
- "Run a health check on device 123, then give me a portal overview for shift handoff"
- "Forecast when memory on device 123 will hit 90%"
- "What's the blast radius if device 789 goes down?"
- "What diagnostics ran on device 123 today, and what did the scripts output?"
The server also takes LogicMonitor filter syntax directly, for example "Get devices where filter is 'displayName~"prod",hostStatus:"normal"'". String values, hostStatus words included, are quoted.
Find a tool
search_tools: keyword search by name and description over the tools your settings list, so with writes off it names no write tool; an optionalcategorynarrows it to one group- the
lm://guide/tool-categoriesresource: all 258 tools grouped by domain - documentation/tools.md: every tool, its parameters, and its read/write classification, generated from the tool registry
What the tools cover
| Area | Tools |
|---|---|
| Alerts | Query, acknowledge, bulk acknowledge and add notes to alerts; full CRUD for alert rules; datapoint thresholds, read at instance, device or group level with the level each one comes from, and set or cleared at instance or group level |
| Devices | Full CRUD for devices and groups, properties, and manual datasource instances |
| Metrics | Datasources, instances, metric data, graphs, and native OTLP metrics through PromQL range and instant queries; analyze_metric(source="otlp") analyzes a PromQL series |
| Dashboards | Full CRUD for dashboards, widgets, and groups |
| SDT | Create, list, bulk create, and bulk delete scheduled downtime |
| Collectors | Collectors and collector groups, each collector's events and uploaded log files, and the upgrade history; with writes on, status checks, restarts and acknowledging a down collector. OpenTelemetry (lmotel) collectors are not among them, since the v3 API the server reads does not list them |
| Websites | Full CRUD for synthetic checks and website groups |
| Reports | List, view, run, and schedule reports |
| Escalations | Full CRUD for escalation chains and recipient groups |
| Users and audit | Users, roles, access groups, API tokens, audit logs, ops notes, login and change audits |
| LogicModules | Query, create, update, delete, and import data, config, event, property, topology, log, diagnostic, and remediation sources, and read any of them as its full REST object with raw=true (remediation sources have no import), plus Config Archive retrieval, diffs, and on-demand collection |
| Diagnostics and remediation | Assigned-source resolution, structured execution results with script output, manual execution, action chains and rules |
| APM traces | Service discovery, per-operation RED metrics, service alerts, and OTel properties |
| Network | Neighbors, interfaces, flows, connections, top talkers, link flaps, and outage detection. See networking intelligence |
| Cost | Cost recommendations, recommendation categories, and idle resources (LM Envision) |
| Ingestion | Push logs and metrics through LMv1 authentication |
| Logs | LM Logs pipelines, tracked queries, saved query groups, partitions with their contracts, and the usage figure. Log search is not offered: the v3 search endpoint applied neither the query nor the time range when probed |
| Batch jobs | List batch jobs and the BatchJob datasources on a device; per-run output comes from get_device_data |
get_network_flows and get_top_talkers read /netflow/flows, which answered 404 on the only portal tested, one with no NetFlow-enabled device. Treat both as unverified until you have tried them on a portal with NetFlow data.
update_logicmodule(type, id, changes, mode) is the safe way to change a LogicModule: it reads the full definition, deep-merges your changes, validates required fields, and returns a dry-run diff by default, so a partial change cannot blank a script.
Analysis tools
These run server-side in pure Python, so an assistant gets the result in one call.
- Alert correlation clusters, statistics by severity and device, noise scoring with flap detection, and burst detection
analyze_metricreads a series once for its summary, trend, seasonality, change points, anomalies (z-score, IQR, MAD) and linear and Holt-Winters forecast to a threshold, and with one more read compares it with an earlier window; metric correlation across series- Availability with MTTR from active and cleared alerts, and SLO error budgets
- Topology-based blast radius, and change correlation against audit logs
Composite tools (triage, diagnose, health_check, capacity_plan, portal_overview) combine these. Each takes detail_level (summary or full) and returns partial results when a sub-tool fails.
Claude Code skills
Slash-command runbooks that ship with the repo. Clone it and invoke /lm-triage in Claude Code.
| Skill | Command | Description |
|---|---|---|
| Alert Triage | /lm-triage |
Investigate active alerts, score noise, correlate clusters, assess blast radius, take action |
| Device Health | /lm-health <device> |
Health check: metrics, anomalies, availability, topology |
| Portal Overview | /lm-portal |
Portal-wide snapshot for shift handoff: alerts, collectors, SDTs, down devices |
| Capacity Planning | /lm-capacity <device> |
Trend analysis, seasonality detection, breach forecasting, right-sizing |
| APM Investigation | /lm-apm [service] |
Service discovery, operation-level RED metrics, alert correlation |
| Remediation | /lm-remediate <alert_id or device> |
Diagnose an alert, then run an assigned RemediationSource or an Edwin AI agent tool after a pre-execution checklist and your confirmation |
| Edwin Review | /lm-edwin [insight] |
Check an Edwin AI insight's root cause against the causal CI's metric and the alert window |
| Logs Review | /lm-logs [partition] |
Review LM Logs partitions and their contracts, the pipelines that name them, usage, and saved and tracked queries |
Integrations
terraform_generate needs no settings. Multi-portal mode and Edwin AI are optional; add their variables to the env block or claude mcp add command from Setup.
Terraform
terraform_generate exports an existing LM portal resource as HCL for the logicmonitor/logicmonitor provider. It is part of the core tool set and needs no settings of its own.
Multi-portal mode
Serves many customer portals from a single server entry. Set LM_MULTI_PORTAL=true and point the server at a credential vault; the full tool set loads once, and four tools (get_portals, use_portal, get_current_portal, run_portal_reload) switch the active portal at runtime. Credentials come from an age-encrypted vault (or a plaintext JSON file for testing), and each portal is read-only unless its record marks it writable. The mode is stdio-only. See MULTIPORTAL.md.
Edwin AI
Set EDWIN_CLIENT_ID and EDWIN_CLIENT_TOKEN to an Edwin AI API credential and the server lists eight more tools. get_edwin_insights and get_edwin_insight read the incidents Edwin built by correlating alerts, with its root cause and remediation analysis when it has written one. get_edwin_records, get_edwin_alert_volume and get_edwin_metadata read Edwin's copy of alerts and events and its data model. get_edwin_tools lists the tools of Edwin's agents and edwin_run_tool runs one; it needs LM_ENABLE_WRITE_OPERATIONS=true, every run is audited, and multi-portal mode does not offer it, since one Edwin tenant serves every portal there. ask_edwin asks Edwin a question and waits up to two minutes for the answer. The server calls https://<portal name>.dexda.ai unless EDWIN_HOST names another host; multi-portal mode has no portal name to derive it from, so set EDWIN_HOST there. A credential that lacks a scope a tool needs gets SCOPE_MISSING naming the scope.
Deployment
Stdio is the default transport and the one MCP clients use. It needs no deployment: your client launches the server.
HTTP transport
The HTTP transport serves health, readiness and info routes only; it exposes no tool and no MCP endpoint, so /mcp and /api/v1/* have no route (404 once past the bearer check). Use stdio for MCP clients.
cd deploy
cp .env.example .env
# Edit .env with your credentials, including LM_HTTP_AUTH_TOKEN (required)
# Run with docker-compose
docker compose up -d
# With TLS via Caddy
docker compose --profile tls up -d
The server exposes health endpoints for container orchestration:
GET /health- Detailed health check with all component statuses (adds the LM API connectivity check when enabled, for callers presentingLM_HTTP_AUTH_TOKENor when no token is set)GET /healthz- Liveness probe (200 OK or 503)GET /readyz- Readiness probe (local checks only; never calls the LM API)
Every route except /healthz and /readyz, /health included, answers only the Host names the server expects (the loopback names, a non-wildcard LM_HTTP_HOST, and LM_HTTP_ALLOWED_HOSTS) and refuses others with 400; those two probes accept any Host so probes by pod IP keep working. The compose file requires LM_HTTP_AUTH_TOKEN for every profile and passes DOMAIN through as an allowed host for the Caddy tls profile.
See SECURITY.md for the exposure model.
Kubernetes
deploy/k8s/deployment.yaml is a vendor-neutral Deployment and Service running
the published ghcr.io image. It works on AKS, EKS, GKE, OpenShift, or vanilla
Kubernetes, reads credentials from a Kubernetes secret you create, and wires the
liveness and readiness probes to the endpoints above. Apply it with
kubectl apply -f deploy/k8s/deployment.yaml after creating the secret
documented at the top of the file.
Configuration
Set these as environment variables, through your client's env block or -e flags. The server does not read a .env file itself; to use one when you run it from a clone, run uv run --env-file .env lm-mcp-server.
Authentication
Set LM_BEARER_TOKEN for most operations. Set LM_ACCESS_ID and LM_ACCESS_KEY as well for ingest_logs and push_metrics, which require LMv1 authentication.
To get LMv1 credentials:
- Go to Settings > Users and Roles > Users.
- Select a user, then open the API Tokens tab.
- Create or view the Access ID and Access Key.
Environment variables
| Variable | Required | Default | Description |
|---|---|---|---|
LM_PORTAL |
Yes | - | LogicMonitor portal hostname (e.g., company.logicmonitor.com) |
LM_BEARER_TOKEN |
Yes* | - | API Bearer token (min 10 characters) |
LM_ACCESS_ID |
No | - | LMv1 API access ID (for ingestion APIs) |
LM_ACCESS_KEY |
No | - | LMv1 API access key (for ingestion APIs) |
LM_ENABLE_WRITE_OPERATIONS |
No | false |
Enable write operations (create, update, delete) |
LM_TIMEOUT |
No | 30 |
Request timeout in seconds (range: 5-300) |
LM_MAX_RETRIES |
No | 3 |
Max retries for rate-limited requests, and for server errors on idempotent requests (GET, PUT, DELETE); a POST or PATCH that gets a 5xx is not resent (range: 0-10) |
LM_RETRY_AFTER_CAP_SECONDS |
No | 120 |
Longest wait between retries. A server's Retry-After can lengthen the backoff up to this cap, never shorten it |
LM_TRANSPORT |
No | stdio |
Transport mode: stdio (MCP clients) or http (see HTTP transport) |
LM_MULTI_PORTAL |
No | false |
Serve many customer portals from one server, selected at runtime via use_portal. Stdio-only. |
LM_VAULT_FILE |
No | - | Path to the age-encrypted portal vault (multi-portal) |
LM_AGE_KEY |
No | - | Path to the age identity that decrypts the vault (multi-portal) |
LM_PORTALS_FILE |
No | - | Plaintext JSON portal map (multi-portal, testing only; the encrypted vault wins when both are set) |
EDWIN_CLIENT_ID |
No | - | Edwin AI API credential client id. With EDWIN_CLIENT_TOKEN it turns on the eight Edwin tools; one without the other leaves them off and logs a warning |
EDWIN_CLIENT_TOKEN |
No | - | Secret of that Edwin AI credential |
EDWIN_HOST |
No | https://<portal name>.dexda.ai |
Edwin AI base URL, https only. Required in multi-portal mode, which has no LM_PORTAL to derive it from |
LM_HTTP_HOST |
No | 0.0.0.0 |
HTTP server bind address. With LM_TRANSPORT=http, an address other than 127.0.0.1, ::1 or localhost needs LM_HTTP_AUTH_TOKEN or LM_HTTP_ALLOW_UNAUTHENTICATED=true, or the server refuses to start. |
LM_HTTP_PORT |
No | 8080 |
HTTP server port |
LM_HTTP_AUTH_TOKEN |
No | - | Require this bearer token on every path outside / and the health endpoints (min 16 chars). Needed to start on a non-loopback LM_HTTP_HOST, the default 0.0.0.0 included, unless LM_HTTP_ALLOW_UNAUTHENTICATED=true. Health endpoints and / stay open for probes. HTTP transport only. |
LM_HTTP_ALLOWED_HOSTS |
No | - | Comma-separated Host names the server accepts on every route except /healthz and /readyz, besides 127.0.0.1, ::1, localhost and a non-wildcard LM_HTTP_HOST (for example the public name a reverse proxy forwards). HTTP transport only. |
LM_HTTP_ALLOW_UNAUTHENTICATED |
No | false |
Start on a non-loopback LM_HTTP_HOST with no LM_HTTP_AUTH_TOKEN, for a listener protected another way (an authenticating proxy, a private network). HTTP transport only. |
LM_HTTP_SSL_CERTFILE |
No | - | TLS certificate path. Set with LM_HTTP_SSL_KEYFILE to serve HTTPS. HTTP transport only. |
LM_HTTP_SSL_KEYFILE |
No | - | TLS private key path. HTTP transport only. |
LM_HTTP_SSL_KEYFILE_PASSWORD |
No | - | Password for an encrypted TLS private key. HTTP transport only. |
LM_SESSION_HISTORY_SIZE |
No | 50 |
Number of tool calls to keep in history |
LM_LOG_LEVEL |
No | warning |
Logging level: debug, info, warning, or error. Records go to stderr; write audit lines (info for a success, warning for a failure) redact argument values under credential-shaped keys such as password, token, api_key and extra_vars, LogicMonitor credential properties such as ssh.pass and snmp.community, and a property_value whose property_name is one |
LM_ENABLED_TOOLS |
No | - | Comma-separated tool names or glob patterns to enable (e.g., get_*,triage). Mutually exclusive with LM_DISABLED_TOOLS. |
LM_DISABLED_TOOLS |
No | - | Comma-separated tool names or glob patterns to disable (e.g., delete_*). Mutually exclusive with LM_ENABLED_TOOLS. |
LM_MCP_CATEGORIES |
No | - | Comma-separated category names to include: read, write, delete, export, import, session, workflow. Composes by intersection with LM_ENABLED_TOOLS/LM_DISABLED_TOOLS -- only narrows, never expands. Useful for clients with tool-count limits (e.g., Cursor's 40-tool cap). |
LM_HEALTH_CHECK_CONNECTIVITY |
No | false |
Include an LM API ping in /health for authenticated callers (never in /readyz) |
*Either LM_BEARER_TOKEN or both LM_ACCESS_ID and LM_ACCESS_KEY are required.
MCP resources and prompts
Clients read these through the standard MCP resource and prompt calls.
| URI | Description |
|---|---|
lm://schema/alerts |
Alert object fields, types, and descriptions |
lm://schema/devices |
Device object fields and types |
lm://schema/sdts |
SDT (Scheduled Downtime) object fields |
lm://schema/dashboards |
Dashboard object fields |
lm://schema/collectors |
Collector object fields |
lm://schema/escalations |
Escalation chain object fields |
lm://schema/reports |
Report object fields |
lm://schema/websites |
Website check object fields |
lm://schema/datasources |
DataSource definition fields |
lm://schema/users |
User object fields |
lm://schema/audit |
Audit log entry fields |
lm://enums/severity |
Alert severity levels: critical(4), error(3), warning(2), info(1) |
lm://enums/device-status |
Device status words: normal, dead, dead-collector, unmonitored, disabled |
lm://enums/sdt-type |
SDT types: DeviceSDT, DeviceGroupSDT, etc. |
lm://enums/alert-cleared |
Alert cleared status: true, false |
lm://enums/alert-acked |
Alert acknowledgment status: true, false |
lm://enums/collector-build |
Collector build types: EA, GD, MGD |
lm://filters/alerts |
Filter fields and operators for alert queries |
lm://filters/devices |
Filter fields and operators for device queries |
lm://filters/sdts |
Filter fields and operators for SDT queries |
lm://syntax/operators |
Filter operators: :, ~, >, <, !:, !~, >:, <: |
lm://guide/tool-categories |
All 258 tools organized by domain category |
lm://guide/examples |
Common filter patterns and query examples |
lm://guide/mcp-orchestration |
Patterns for combining LogicMonitor with other MCP servers |
lm://guide/best-practices |
Scenario-based best practices with recommendations and anti-patterns |
| Prompt | Description | Arguments |
|---|---|---|
incident_triage |
Analyze active alerts, identify patterns, suggest root cause | severity, time_window_hours |
capacity_review |
Review resource utilization and identify capacity concerns | group_id, threshold_percent |
remediation |
Execute a LogicMonitor remediation source with pre-execution safety checks | host_id, remediation_source_id, alert_id |
Troubleshooting
The server logs to stderr. Set LM_LOG_LEVEL=debug to add httpx and MCP SDK lines.
"Failed to connect" in Claude Code
claude mcp list shows this when the server is missing environment variables. Claude Code spawns the server as a subprocess with its own environment, so variables in your shell or a .env file do not reach it, and the -e flags are required:
# Remove the broken config
claude mcp remove logicmonitor
# Re-add with environment variables
claude mcp add logicmonitor \
-e LM_PORTAL=yourcompany.logicmonitor.com \
-e LM_BEARER_TOKEN=<your-bearer-token> \
-- uvx --from lm-mcp lm-mcp-server
"Write operations are disabled"
Write operations (acknowledge, create SDT, etc.) are disabled by default: the server does not list those tools, and a call to one by name returns this error. Set LM_ENABLE_WRITE_OPERATIONS=true in your environment.
LogicMonitor MCP questions
How do I add LogicMonitor MCP to Claude Code?
Run claude mcp add --env 'LM_BEARER_TOKEN=<YOUR_LM_BEARER_TOKEN>' --env 'LM_PORTAL=<LM_PORTAL>' --env 'EDWIN_CLIENT_TOKEN=<YOUR_EDWIN_CLIENT_TOKEN>' --env 'LM_ACCESS_KEY=<YOUR_LM_ACCESS_KEY>' --env 'LM_HTTP_AUTH_TOKEN=<YOUR_LM_HTTP_AUTH_TOKEN>' --env 'LM_HTTP_SSL_KEYFILE_PASSWORD=<YOUR_LM_HTTP_SSL_KEYFILE_PASSWORD>' mcp-server-logicmonitor -- uvx lm-mcp in your terminal. Add --scope user to make it available in every project.
How do I add LogicMonitor MCP to Cursor?
Open ~/.cursor/mcp.json and add the server under mcpServers, using the Cursor config shown in the install section above.
What tools does LogicMonitor MCP provide?
It exposes 1 tool, including search_tools.
Can the AI assistant make changes to my LogicMonitor portal?
By default, the server only lists read-only tools. You must set the LM_ENABLE_WRITE_OPERATIONS environment variable to true to enable tools that change your portal.
What kind of analysis can the server perform?
The server can perform alert correlation, noise scoring, metric forecasting, availability calculations, and topology-based blast radius analysis.
Does it support Terraform?
Yes, it includes a terraform_generate tool that exports existing LogicMonitor resources as HCL for the logicmonitor provider.