Axiom MCP Server

Provider-hosted MCP server for Axiom's observability platform. Query datasets with APL, inspect schemas, monitor alerts, and manage dashboards.

Observability by Axiom OAuth2 active

Connect (Remote Server) GitHub Docs

Overview

The Axiom MCP server connects AI assistants to Axiom's observability platform, allowing agents to query logs, traces, metrics, and events using Axiom Processing Language (APL). It is hosted by Axiom at https://mcp.axiom.co/mcp, so there are no local dependencies. Authentication happens through the browser via OAuth and can be revoked from the Axiom Console.

The server exposes a comprehensive set of read and write tools for observability workflows: schema discovery, APL and metrics queries, saved query retrieval, monitor inspection with history, and full dashboard management (create, read, update, delete, export, chart updates). It also ships a set of prompts for common workflows like anomaly detection, data quality investigation, cross-dataset correlation, performance baselining, dataset exploration, and monitor health analysis.

Axiom's role-based access control applies to MCP sessions, so agents only see data the calling user is permitted to access, and every query is recorded in Axiom's audit trail. A self-host option is available via the open-source Go binary at github.com/axiomhq/axiom-mcp for users who want to run the server locally.

Tools

Tool	Description
`queryApl`	Execute an APL (Axiom Processing Language) query against an Axiom dataset.
`listDatasets`	List available Axiom datasets in the organization.
`getDatasetSchema`	Return the schema (fields and types) for a specified dataset.
`getSavedQueries`	Retrieve saved/starred APL queries from the workspace.
`getMonitors`	List current monitors configured in Axiom.
`getMonitorsHistory`	Return historical firing data for monitors to spot trends and anomalies.
`queryMetrics`	Run a metrics query against Axiom metric data.
`searchMetrics`	Search for metrics by name or attributes.
`listMetrics`	List available metrics.
`listMetricTags`	List tags associated with a metric.
`getMetricTagValues`	Return values for a specific metric tag.
`listDashboards`	List dashboards in the workspace.
`getDashboard`	Fetch a dashboard definition.
`createDashboard`	Create a new dashboard.
`updateDashboard`	Update an existing dashboard.
`updateDashboardChart`	Update a specific chart on a dashboard.
`deleteDashboard`	Delete a dashboard.
`exportDashboard`	Export a dashboard definition.

Setup Guide

Prerequisites

An Axiom account (Pro, Team, or Enterprise plan for the remote server)
At least one dataset with ingested data

Remote server (recommended)

The server is hosted at https://mcp.axiom.co/mcp. Authentication is handled via browser-based OAuth.

Claude Desktop / Cursor / generic MCP client

{
  "mcpServers": {
    "axiom": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://mcp.axiom.co/mcp"]
    }
  }
}

For clients with native remote MCP support, point directly at https://mcp.axiom.co/mcp (or https://mcp.axiom.co/sse for SSE-based agents) and complete the browser OAuth flow.

Header-based authentication (for agents without OAuth)

Send a personal access token as a Bearer token plus your organization ID:

Authorization: Bearer PERSONAL_ACCESS_TOKEN
x-axiom-org-id: ORGANIZATION_ID

Or pass the org ID as a query parameter: https://mcp.axiom.co/mcp?org-id=ORGANIZATION_ID.

Self-hosted (optional)

Install the Go binary:

go install github.com/axiomhq/axiom-mcp@latest

Then point your MCP client at the local binary with your Axiom API token configured.

Use Cases

Ask an agent to investigate a production incident by running APL queries across logs and traces and correlating events across datasets.
Have an agent enumerate datasets, fetch schemas, and propose useful APL queries for a new team member exploring the data.
Review monitor history to detect flapping alerts or identify which monitors triggered during a recent outage window.
Build or update Axiom dashboards from natural language, including chart-level edits and exports.
Run metrics queries and tag drill-downs to baseline performance or hunt anomalies without writing APL by hand.

Example Prompts

"List all datasets, then show the schema of the http-logs dataset."
"Run an APL query to find the top 10 status 500 endpoints in the last 24 hours from http-logs."
"Which monitors fired most often this week? Pull their history and summarize the patterns."
"Search metrics for anything related to checkout latency and show me p95 over the last 6 hours."
"Create a dashboard called 'Payments Health' with charts for error rate, p95 latency, and request volume from the payments dataset."

Pros

Officially hosted by Axiom with zero local setup and browser-based OAuth.
Broad tool coverage: APL, metrics, monitors with history, saved queries, and full dashboard CRUD.
Inherits Axiom RBAC and audit logging, so agent access is scoped and traceable.
Ships with prebuilt MCP prompts for common observability workflows.

Limitations

Remote server requires an Axiom Pro, Team, or Enterprise plan.
Read and dashboard write tools are present, but ingestion and admin operations remain outside MCP scope.
Useful queries assume familiarity with APL or careful prompting to let the agent author it.

Alternatives

Grafana MCP server for querying Grafana, Loki, Prometheus, and Tempo.
Honeycomb MCP server for similar observability queries against Honeycomb.
Sentry MCP server for error and performance monitoring data.