Query Prometheus Metrics with LLMs

Prometheus MCP Server

About

This is an MCP server to allow LLMs to interact with a running Prometheus instance via the API to do things like generate and execute promql queries, list and analyze metrics, etc.

Demos and Examples

Asking Claude to Investigate Slow Queries

The prompt used was:

querying my metrics is slow, can you help me figure out why?

Investigate metrics produced by the MCP server itself and suggest recording rules for SLOs

The prompt used was:

use the tools from the prometheus mcp server to investigate the metrics from
the mcp server and suggest prometheus recording rules for SLOs

Summarize Prometheus metric/label naming best practices

The prompt used was:

summarize prometheus metric/label name best practices

Report on the health of the Prometheus instance that powers prometheus.demo.prometheus.io

The prompt used was:

please provide a comprehensive review and summary of the prometheus server.
review it's configuration, flags, runtime/build info, and anything else that
you feel may provide insight into the status of the prometheus instance,
including analyzing metrics and executing queries

Tools

The Prometheus HTTP API outputs JSON data, and the tools in this MCP server return that JSON to the LLM for processing as it's structured and well understood by LLMs.

LLMs and Token/Context Efficiency

This MCP server supports the following options which have the potential to reduce token/context usage:

TOON Encoding

If token/context usage is a concern, this MCP server also supports converting the API's JSON data to the Token-Oriented Object Notation (TOON) format.
While it is not guaranteed to reduce token usage, it is designed with token efficiency in mind.
As noted on TOON's documentation, it excels at uniform arrays of objects; non-uniform/complex objects may still be more token-efficient in JSON.
Real world token usage will depend on usage patterns, please review common workflows to determine if TOON output may be beneficial.
Please see Flags for more information on the available flags and their corresponding environment variables.

API Response Truncation

This feature allows you to set a maximum limit on the number of lines or entries returned from the Prometheus API for, which can help in reducing the amount of data sent to the LLM.
Setting the limit to 0 disables truncation.
Truncation is disabled by default.
Note that LLMs capable of handling tool request arguments can override this global truncation limit on a per-tool-call basis for supported tools.
Please see Flags for more information on the available flags and their corresponding environment variables.

Full Tool List

NOTE:

Because the TSDB Admin API endpoints
allow for potentially destructive operations like deleting data, they are not
enabled by default. In order to enable the TSDB Admin API endpoints, the MCP
server must be started with the flag --dangerous.enable-tsdb-admin-tools to
acknowledge the associated risk these endpoints carry.

Tool Sets

The server exposes many tools to interact with Prometheus. There are tools to interact with Prometheus via the API, as well as additional tools to do things like read documentation, etc.
By default, they are all registered and available for use (TSDB Admin API tools need an extra flag).

To be considerate to LLMs with smaller context windows, it's possible to pass in a whitelist of specific tools to register with the server.
The following 'core' tools are always loaded: [docs_list, docs_read, docs_search, runbooks_list, runbooks_read, query, range_query, metric_metadata, label_names, label_values, series].
Additional tools can be specified with the --mcp.tools flag.

For example, the command line:

prometheus-mcp-server --mcp.tools=build_info --mcp.tools=flags --mcp.tools=runtime_info

Would result in the following tools being loaded:

• build_info
• docs_list
• docs_read
• docs_search
• flags
• label_names
• label_values
• metric_metadata
• query
• range_query
• runbooks_list
• runbooks_read
• runtime_info
• series

Runbooks (Agent Skills) and Prompts

The server embeds a set of runbooks: guided workflows for common Prometheus tasks, expressed in terms of the server's tools. Each runbook orients the model on the relevant tools, then suggests topics to explore with example queries rather than prescribing a fixed sequence of steps.
Runbooks cover tasks like system health checks, missing-data triage, error-rate investigation, high-cardinality optimization, recording/alerting rule review, and configuration/performance tuning.

Each runbook is packaged as a full Agent Skill: a directory containing a SKILL.md with name/description frontmatter.
Runbooks are exposed three ways:

• Tools: the model can discover and read them itself via the runbooks_list/runbooks_read tools when a request matches a runbook's purpose. Tools are the most portable path and work in every MCP client.
• Skill resources: per the SEP-2640 skills extension draft, each runbook is a skill://<name>/SKILL.md resource, enumerated by the well-known skill://index.json discovery index, and the server declares the io.modelcontextprotocol/skills extension capability. Skill-aware hosts can consume these like local filesystem skills; other clients can still read them as ordinary MCP resources.
• MCP prompts: each runbook is also registered as an MCP prompt under its skill name (e.g. check-system-health, optimize-high-cardinality), so clients with prompt support can invoke a guided workflow directly (often surfaced as slash commands).

Prometheus Compatible Backends

There are many Prometheus compatible backends that can be used to extend prometheus in a variety of ways, often with the goals of offering long term storage or query aggregation from multiple prometheus instances.
Some examples can be found in the Remote Storage of prometheus' docs.

Many of those services also offer a "prometheus compatible" API that can be used to query/interact with the data using native promQL.
In general, this MCP server should at a minimum work for other prometheus API compatible services to execute queries and interact with the series/labels/metadata endpoints for metric and label discovery.
Beyond that, there may be API differences as the different systems implement different parts/extensions of the API for their needs.

Examples:

• Thanos does not use a centralized config, so the config endpoint is not implemented and thus the config tool fails.
• Mimir and Cortex implement extra endpoints to manage/add/remove rules

To workaround this and provide a better experience on some of the commonly used Prometheus compatible systems, this project may add direct support for select systems to provide different/more tools.
Choosing a specific prometheus backend implementation can be done with the --prometheus.backend flag.
The list of available backend implementations on a given release of the MCP server can be found in the output of the --help flag.
Qualifications and support criteria are still under consideration, please open an issue to request support/features for a specific backend for further discussion.

Prometheus Backend Implementation Differences

Resources

Installation and Usage

This MCP server is most useful when fully integrated with tooling and/or installed as a tool server with another system.
Installation procedures and integration support will vary depending on the tools being used.
For example:

• some systems can only interact with MCP tools and not resources/prompts
• some systems use mcp.json config file format to manage MCP servers and some require custom formats
• some systems don't speak MCP directly and require tools like mcp-to-openapi to proxy

Please check the documentation for the tool being used/integrated for specific instructions and level of support.

Binary

Download a release appropriate for your system from the Releases page.
Please see Flags for more information on the available flags and their corresponding environment variables.

/path/to/prometheus-mcp-server <flags>

# or using env vars
PROMETHEUS_MCP_SERVER_PROMETHEUS_URL="https://$yourPrometheus:9090" /path/to/prometheus-mcp-server

Docker

Please see Flags for more information on the available flags and their corresponding environment variables.

# Stdio transport
docker run --rm -i ghcr.io/tjhop/prometheus-mcp-server:latest --prometheus.url "https://$yourPrometheus:9090" 

# or using env vars
docker run --rm -i -e PROMETHEUS_MCP_SERVER_PROMETHEUS_URL="https://$yourPrometheus:9090" ghcr.io/tjhop/prometheus-mcp-server:latest

# Streamable HTTP transport (capable of SSE as well)
docker run --rm -p 8080:8080 ghcr.io/tjhop/prometheus-mcp-server:latest --prometheus.url "https://$yourPrometheus:9090" --mcp.transport "http" --web.listen-address ":8080"

# or using env vars
docker run --rm -p 8080:8080 -e PROMETHEUS_MCP_SERVER_PROMETHEUS_URL="https://$yourPrometheus:9090" -e PROMETHEUS_MCP_SERVER_MCP_TRANSPORT="http" -e PROMETHEUS_MCP_SERVER_WEB_LISTEN_ADDRESS=":8080" ghcr.io/tjhop/prometheus-mcp-server:latest

Helm Chart (Kubernetes)

A Helm chart is available for deploying to Kubernetes. The chart is published as an OCI artifact on each release.

helm install prometheus-mcp-server oci://ghcr.io/tjhop/charts/prometheus-mcp-server \
  --version <version> \
  --set prometheus.url=http://prometheus:9090

See the chart README for the full list of configuration values and examples.

Kubernetes (Raw Manifests)

If you prefer deploying without Helm, example Kubernetes manifests are available:

• Kubernetes Deployment
• OpenShift Deployment

System Packages

Download a release appropriate for your system from the Releases page. A Systemd service file is included in the system packages that are built.

# install system package (example assuming Debian based)
apt install /path/to/package
# create unit override, add any needed flags or environment variables
systemctl edit prometheus-mcp-server.service
systemctl enable --now prometheus-mcp-server.service

Note: While packages are built for several systems, there are currently no plans to attempt to submit packages to upstream package repositories.

Security and Authentication

Connecting to Secure Prometheus Instances

The MCP server supports Prometheus HTTP config files to connect to secured Prometheus instances.
An example config can be found in the examples folder here.
Use the --http.config command-line flag to provide an HTTP configuration file.
Please see Flags for more information.

Securing the MCP Server Endpoints

The MCP server supports Prometheus Web Configuration files files to expose it's endpoints behind optional basic auth and custom TLS configs.
Use the --web.config.file command-line flag to provide an HTTP configuration file.
Please see Flags for more information.

Telemetry

Metrics

Once running, the server exposes Prometheus metrics on the configured listen address and telemetry path (:8080/metrics, by default).
Please see Flags for more information on how to change the listening interface, port, or telemetry path.

Grafana Dashboard

A pre-built Grafana dashboard is included in the grafana/ directory for visualizing the metrics exposed by the MCP server. Import the dashboard json into grafana and it should be ready to go.

Logs

This project makes heavy use of structured, leveled logging.
Please see Flags for more information on how to set the log format, level, and optional file.

Development

Development Environment with Devbox + Direnv

If you use Devbox and
Direnv, then simply entering the directory for the repo
should set up the needed software.

Local LLM with Ollama

See mcp.json for an example MCP config for use with tooling.
Requires ollama to be installed.

NOTE:

To override the default LLM (ollama:gpt-oss:20b), run export OLLAMA_MODEL="ollama:your_model" to override it before running make .

make inspector

make mcphost

Gemini with gemini-cli

See settings.json for an example config file to run gemini-cli with the prometheus-mcp-server.
Requires gemini-cli to be installed.

make gemini

Manual Setup

Required software:

• Working Go environment
• Docker for local tests/linting/image building/etc
• GNU Make
• ollama
• mcp inspector
• mcphost

Building

This project uses the standard Prometheus build tooling: promu
driven through Makefile / Makefile.common. The binary embeds a pinned
snapshot of prometheus/docs, which the build targets download and extract
automatically (the pin lives in DOCS_VERSION in the Makefile).

make build       # build the binary for the host platform (via promu)
make test        # run the test suite
make crossbuild  # build binaries for all release platforms
make             # run the full check suite: style, license, yamllint, lint, build, test

Project-specific helper targets (helm packaging and the local LLM client
integrations) are listed by make help:

make help

Usage:
  make <target>

Project targets:
  help                           print this help message (see Makefile.common for the standard prometheus targets)
  docs                           download and extract the pinned prometheus/docs snapshot for embedding
  helm-sync-dashboards           copy grafana dashboards into helm chart for packaging
  helm-lint                      run helm chart linting
  helm-template                  render helm templates for inspection
  helm-test                      install helm chart and run tests (requires a running cluster)
  mcphost                        use mcphost to run the prometheus-mcp-server against a local ollama model
  inspector                      use inspector to run the prometheus-mcp-server in STDIO transport mode
  inspector-http                 use inspector to run the prometheus-mcp-server in streamable HTTP transport mode
  open-webui                     use open-webui to run the prometheus-mcp-server
  gemini                         use gemini-cli to run the prometheus-mcp-server against Google Gemini models

Command Line Flags

The available command line flags are documented in the help flag:

~/go/src/github.com/tjhop/prometheus-mcp-server (main [ ]) -> ./prometheus-mcp-server --help
usage: prometheus-mcp-server [<flags>]


Flags:
  -h, --[no-]help                Show context-sensitive help (also
                                 try --help-long and --help-man).
                                 ($PROMETHEUS_MCP_SERVER_HELP)
      --mcp.tools=all ...        List of mcp tools to load. The target
                                 `all` can be used to load all tools.
                                 The target `core` loads only the core tools:
                                 docs_list,docs_read,docs_search,runbooks_list,runbooks_read,query,range_query,metric_metadata,label_names,label_values,series
                                 Otherwise, it is treated as an allow-list
                                 of tools to load, in addition to the core
                                 tools. Please see project README for more
                                 information and the full list of tools.
                                 ($PROMETHEUS_MCP_SERVER_MCP_TOOLS)
      --[no-]mcp.enable-toon-output  
                                 Enable Token-Oriented Object Notation
                                 (TOON) output for tools instead of JSON
                                 ($PROMETHEUS_MCP_SERVER_MCP_ENABLE_TOON_OUTPUT)
      --[no-]mcp.enable-client-logging  
                                 Enable sending log messages to connected
                                 MCP clients as protocol notifications.
                                 When enabled, tool execution logs are
                                 sent both to the server's primary
                                 log output and to the MCP client,
                                 allowing LLMs to observe server activity.
                                 ($PROMETHEUS_MCP_SERVER_MCP_ENABLE_CLIENT_LOGGING)
      --mcp.transport="stdio"    The type of transport to use for
                                 the MCP server [`stdio`, `http`].
                                 ($PROMETHEUS_MCP_SERVER_MCP_TRANSPORT)
      --prometheus.backend=PROMETHEUS.BACKEND  
                                 Customize the toolset for a specific
                                 Prometheus API compatible backend.
                                 Supported backends include: prometheus,thanos
                                 ($PROMETHEUS_MCP_SERVER_PROMETHEUS_BACKEND)
      --prometheus.url="http://127.0.0.1:9090"  
                                 URL of the Prometheus instance to connect to
                                 ($PROMETHEUS_MCP_SERVER_PROMETHEUS_URL)
      --prometheus.timeout=1m    Timeout for API calls to the Prometheus backend
                                 ($PROMETHEUS_MCP_SERVER_PROMETHEUS_TIMEOUT)
      --prometheus.truncation-limit=0  
                                 If enabled, this controls the maximum query
                                 response size in number of lines/entries
                                 provided to the LLM from the API response.
                                 LLMs can override truncation limits if
                                 needed on a per-tool-call basis via tool
                                 request arguments on supported tools.
                                 To disable truncation limits, set to 0.
                                 ($PROMETHEUS_MCP_SERVER_PROMETHEUS_TRUNCATION_LIMIT)
      --http.config=HTTP.CONFIG  Path to config file to set
                                 Prometheus HTTP client options
                                 ($PROMETHEUS_MCP_SERVER_HTTP_CONFIG)
      --web.telemetry-path="/metrics"  
                                 Path under which to expose metrics.
                                 ($PROMETHEUS_MCP_SERVER_WEB_TELEMETRY_PATH)
      --web.max-requests=40      Maximum number of parallel scrape
                                 requests. Use 0 to disable.
                                 ($PROMETHEUS_MCP_SERVER_WEB_MAX_REQUESTS)
      --[no-]dangerous.enable-tsdb-admin-tools  
                                 Enable and allow using tools that access
                                 Prometheus' TSDB Admin API endpoints
                                 (`snapshot`, `delete_series`, and
                                 `clean_tombstones` tools). This is dangerous,
                                 and allows for destructive operations
                                 like deleting data. It is not the fault
                                 of this MCP server if the LLM you're
                                 connected to nukes all your data. Docs:
                                 https://prometheus.io/docs/prometheus/latest/querying/api/#tsdb-admin-apis
                                 ($PROMETHEUS_MCP_SERVER_DANGEROUS_ENABLE_TSDB_ADMIN_TOOLS)
      --[no-]docs.auto-update    Enable automatic documentation updates
                                 from the official prometheus/docs
                                 repository. Checks every 24h0m0s.
                                 ($PROMETHEUS_MCP_SERVER_DOCS_AUTO_UPDATE)
      --log.file=LOG.FILE        The name of the file to log to (file
                                 rotation policies should be configured
                                 with external tools like logrotate)
                                 ($PROMETHEUS_MCP_SERVER_LOG_FILE)
      --[no-]web.systemd-socket  Use systemd socket activation listeners
                                 instead of port listeners (Linux only).
                                 ($PROMETHEUS_MCP_SERVER_WEB_SYSTEMD_SOCKET)
      --web.listen-address=:8080 ...  
                                 Addresses on which to expose metrics and
                                 web interface. Repeatable for multiple
                                 addresses. Examples: `:9100` or `[::1]:9100`
                                 for http, `vsock://:9100` for vsock
                                 ($PROMETHEUS_MCP_SERVER_WEB_LISTEN_ADDRESS)
      --web.config.file=""       Path to configuration file that can
                                 enable TLS or authentication. See:
                                 https://github.com/prometheus/exporter-toolkit/blob/master/docs/web-configuration.md
                                 ($PROMETHEUS_MCP_SERVER_WEB_CONFIG_FILE)
      --log.level=info           Only log messages with the given severity
                                 or above. One of: [debug, info, warn, error]
                                 ($PROMETHEUS_MCP_SERVER_LOG_LEVEL)
      --log.format=logfmt        Output format of log messages. One of: [logfmt,
                                 json] ($PROMETHEUS_MCP_SERVER_LOG_FORMAT)
      --[no-]version             Show application version.
                                 ($PROMETHEUS_MCP_SERVER_VERSION)