SigNoz MCP server: query logs, traces and metrics from your AI assistant

• Finds log lines that match what you are looking for

  Searches your logs by service, level, text or time and returns the matching log lines.

  Try: “Show me error logs for the paymentservice from the last hour”

  Read · signoz_search_logs

• Finds traces and spans for a service or request

  Searches spans or whole traces, including AI traces, by service, time and attributes, and can find trace IDs.

  Try: “Search traces for the checkout service from the last hour”

  Read · signoz_search_traces

• Opens one trace with every span in order

  Shows a single trace with all of its spans and how they nest, so you can see where time went.

  Try: “Get details for trace ID abc123”

  Read · signoz_get_trace_details

• Pulls values and trends for a metric

  Queries a metric you name for its values over time, breakdowns by label, or formulas across metrics.

  Try: “What's the p99 latency for http_request_duration_seconds?”

  Read · signoz_query_metrics

• Lists alerts that are firing or silenced right now

  Lists alert instances that are currently firing, silenced or inhibited.

  Try: “List all active alerts”

  Read · signoz_list_alerts

• Lists the services sending traces

  Lists the services that sent traces in a time range.

  Try: “Show me all available services”

  Read · signoz_list_services

• Counts and groups logs, like errors per service

  Calculates counts, averages or percentiles over logs and groups them, such as errors per service.

  Try: “How many errors per service in the last hour?”

  Read · signoz_aggregate_logs

• Summarises trace stats, like slowest endpoints

  Calculates statistics over spans or AI traces, such as request counts or p99 latency, grouped by any field.

  Try: “Which endpoints had the highest p99 latency today?”

  Read · signoz_aggregate_traces

• Shows the busiest or slowest operations for a service

  Ranks the operations inside one traced service, so you can spot the slow or busy ones.

  Try: “What are the top operations for the cart service?”

  Read · signoz_get_service_top_operations

• Gives an overview of your SigNoz setup and its health

  Reports the current state of your SigNoz deployment with its main usage statistics.

  Try: “Give me an overview of our SigNoz setup”

  Read · signoz_get_org_overview

• Lists the metrics you collect

  Lists active metric names with their type and unit, so the assistant knows what it can query.

  Try: “Show me all available metrics”

  Read · signoz_list_metrics

• Shows which metrics take up the most volume

  Ranks the top 100 metrics by how many samples they ingest, with percentages, to help cut cost.

  Try: “Which metrics are driving most of our ingestion volume?”

  Read · signoz_get_top_metrics

• Shows which dashboards and alerts use a metric

  Checks up to 50 metric names and tells you which dashboards and alerts depend on each one.

  Try: “Is any dashboard or alert still using http_requests_legacy?”

  Read · signoz_check_metric_usage

• Shows which labels on a metric have the most values

  Lists a metric's labels with how many distinct values each has, highest first, to find cardinality problems.

  Try: “Which labels make http_server_duration so expensive?”

  Read · signoz_check_metric_cardinality

• Lists your alert rules, including disabled ones

  Lists every configured alert rule with a summary, including ones that are quiet or turned off.

  Try: “List all our alert rules”

  Read · signoz_list_alert_rules

• Opens the full definition of one alert rule

  Shows the complete setup of one alert rule, including its query and thresholds.

  Try: “Show me the definition of the high error rate alert”

  Read · signoz_get_alert

• Shows when an alert fired and changed state

  Shows the firing and state-change history of one alert rule.

  Try: “When did the checkout latency alert fire this week?”

  Read · signoz_get_alert_history

• Creates a new alert rule

  Creates a threshold, PromQL or anomaly alert rule and sets where it sends notifications.

  Try: “Create an alert when checkout error rate goes above 5% for 5 minutes”

  Write · signoz_create_alert

• Changes an existing alert rule

  Replaces an existing alert rule with an edited version.

  Try: “Raise the threshold on the high latency alert to 800 ms”

  Write · signoz_update_alert

• Deletes an alert rule

  Permanently deletes an alert rule after you confirm it.

  Try: “Delete the old staging disk alert”

  Delete · signoz_delete_alert

• Lists your dashboards

  Lists your dashboards with short summaries.

  Try: “What dashboards do we have?”

  Read · signoz_list_dashboards

• Opens a dashboard with its panels and queries

  Shows one dashboard's full layout, variables, panels and the queries behind them.

  Try: “Open the API latency dashboard”

  Read · signoz_get_dashboard

• Creates a new dashboard

  Builds a new custom dashboard with the panels you describe.

  Try: “Create a dashboard with CPU and memory widgets”

  Write · signoz_create_dashboard

• Rewrites an existing dashboard

  Replaces a dashboard with an edited version while keeping the parts you did not ask to change.

  Try: “Rename the panels on the API dashboard to plain titles”

  Write · signoz_update_dashboard

• Makes a small change to a dashboard

  Applies a targeted edit to a dashboard, such as adding one panel, without resending the whole thing.

  Try: “Add a p95 latency panel to the checkout dashboard”

  Write · signoz_patch_dashboard

• Deletes a dashboard

  Permanently deletes a dashboard after you confirm it.

  Try: “Delete the old test dashboard”

  Delete · signoz_delete_dashboard

• Finds ready-made dashboard templates

  Searches SigNoz's catalog of ready-made dashboards so one can be imported.

  Try: “Is there a ready-made dashboard for PostgreSQL?”

  Read · signoz_list_dashboard_templates

• Adds a ready-made dashboard from a template

  Creates a dashboard from one of SigNoz's curated templates.

  Try: “Import the Kubernetes cluster dashboard template”

  Write · signoz_import_dashboard

• Lists saved Explorer views

  Lists saved views for logs, traces, metrics or Cost Meter.

  Try: “What saved log views do we have?”

  Read · signoz_list_views

• Opens one saved view

  Shows the complete definition of one saved Explorer view.

  Try: “Open the saved view for payment errors”

  Read · signoz_get_view

• Saves a search as a reusable view

  Saves a logs, traces or metrics query as a view you can reopen later.

  Try: “Save this error search as a view called Payment errors”

  Write · signoz_create_view

• Changes a saved view

  Replaces a saved view with an edited version while keeping the rest.

  Try: “Add the region filter to the Payment errors view”

  Write · signoz_update_view

• Deletes a saved view

  Permanently deletes a saved view after you confirm it.

  Try: “Delete the old checkout debug view”

  Delete · signoz_delete_view

• Lists where alerts get sent

  Lists notification channels, such as Slack or email, by name without their settings.

  Try: “Where do our alerts get sent?”

  Read · signoz_list_notification_channels

• Opens one channel's settings

  Shows all provider settings for one notification channel.

  Try: “Show the settings for the on-call Slack channel”

  Read · signoz_get_notification_channel

• Adds a new place to send alerts

  Creates a notification channel and can send a test message if you ask.

  Try: “Add a Slack channel for the payments team alerts”

  Write · signoz_create_notification_channel

• Changes a channel's settings

  Replaces a notification channel's settings and can send a test message if you ask.

  Try: “Point the on-call channel at the new Slack webhook”

  Write · signoz_update_notification_channel

• Deletes a notification channel

  Permanently deletes a notification channel after you confirm it.

  Try: “Delete the old email channel”

  Delete · signoz_delete_notification_channel

• Shows prices for AI models, and models with no price

  Lists the AI model prices SigNoz uses, or models seen in the last hour that have no price set.

  Try: “Which AI models we call have no price set?”

  Read · signoz_get_ai_pricing

• Lists how AI span details are mapped

  Lists the rules that turn vendor-specific AI span details into standard fields.

  Try: “Show our AI attribute mappings”

  Read · signoz_list_ai_attribute_mappings

• Tests AI span mappings without saving

  Runs sample spans through the AI attribute mappings to show the result, without saving anything.

  Try: “Test how these sample spans would be mapped”

  Read · signoz_test_ai_attribute_mapping

• Searches SigNoz's documentation

  Searches the official SigNoz docs for setup, instrumentation and troubleshooting answers.

  Try: “How do I send Docker logs to SigNoz?”

  Network · signoz_search_docs

• Reads one SigNoz docs page

  Fetches one official SigNoz docs page or section as text.

  Try: “Read the SigNoz page on log pipelines”

  Network · signoz_fetch_doc

• Used by the assistant behind the scenes

  Lists the fields you can filter on

  Lists the field names available in metrics, traces or logs, so the assistant can build a correct filter.

  Try: “What fields can I filter logs by?”

  Read · signoz_get_field_keys

• Lists the values a field can have

  Lists the possible values for one field, such as every service name.

  Try: “What values does deployment.environment have?”

  Read · signoz_get_field_values

• Runs an advanced query the other tools cannot express

  Runs a raw SigNoz query, including formulas, PromQL or ClickHouse SQL, when no simpler tool fits.

  Try: “Run a PromQL query for request rate by status code”

  Read · signoz_execute_builder_query

Watch out for

• It can delete dashboards, alerts and channels

  Delete tools exist for alert rules, dashboards, saved views and notification channels. Use a service account with the viewer role unless you want the assistant to change things. github.com

• Your logs and traces go to the AI model

  Log lines and span details the assistant reads are sent to your AI provider. Keep secrets out of logs and ask scoped questions. signoz.io

• A shared self-run server with one key is open

  SigNoz warns against exposing a self-hosted web server set up with one shared key to untrusted callers. For local use, listen on 127.0.0.1 only. signoz.io

• Never commit a config file with your key

  Header sign-in and self-hosted setups put the API key in a config file. Keep that file out of version control. signoz.io

• Channel changes can send a test alert

  Creating or updating a notification channel can send a test message, but only when asked; the default is no test. signoz.io

Security advisories

In their words