Open-source SigNoz MCP alternatives
SigNoz is open source, and so is its MCP server, which is part of the appeal for an OpenTelemetry-native stack you want to read and run yourself. If you are comparing it against other observability servers on that basis, the ones below all publish their code, so you can audit exactly which API calls an agent can make before you grant it access to your telemetry.
Ordered roughly by how widely each is used, these range from full Grafana-style suites to a single-signal Prometheus server. Most install locally; a few also offer a hosted endpoint. All of them let you pin the version you trust.
The 8 best open-source alternatives
Grafana Labs publishes its server, which queries dashboards, Prometheus, and Loki, plus incidents, alerts, and OnCall. Reading the repo tells you precisely which data sources and panels an agent can reach.
Set up Grafana →Sentry's server is open source and pulls issues, stack traces, and events, with Seer root-cause analysis. You can see how it scopes access to a project before it ever touches your error stream.
Set up Sentry →This community Prometheus server is fully open and focused on metrics: PromQL instant and range queries, metric and metadata discovery, and scrape-target inspection, all in a repo you can read end to end.
Set up Prometheus →Read-only by default and open source, the PagerDuty server exposes incidents, services, schedules, teams, and orchestrations across 64 tools, so you can confirm it won't mutate on-call state without explicit opt-in.
Set up PagerDuty →- DatadogOfficial
Open client, hosted backend: the Datadog server publishes its source while searching logs, querying metrics, pulling APM traces, and investigating incidents, so you audit the connector even though the data lives in Datadog.
Set up Datadog → - HoneycombOfficial
Open source and built for high-cardinality work, Honeycomb's server queries traces, metrics, and logs, runs BubbleUp, and manages Boards, Triggers, and SLOs, with the code there to inspect.
Set up Honeycomb → - New RelicOfficial
New Relic's hosted server is open source: it runs NRQL, searches entities, analyzes golden metrics, and triages incidents. You read the connector even though queries execute against New Relic's platform.
Set up New Relic → Broader than pure observability, AWS Labs' open-source server runs any AWS CLI command with validation and a read-only mode, which reaches CloudWatch and the rest of the account if your signals live in AWS.
Set up AWS (AWS Labs) →
How to choose
Every option here ships its source, so the real choice is scope. Grafana is the closest open-source suite to SigNoz across all three signals; Prometheus is the tightest if you only query metrics. Sentry and Honeycomb lean into errors and high-cardinality debugging, while Datadog and New Relic publish open clients to hosted backends. Whichever you pick, read the repo before wiring it to production telemetry with write access.
FAQ
- Is the SigNoz MCP server open source?
- Yes. SigNoz is an open-source, OpenTelemetry-native stack, and its MCP server ships publicly, so you can read the code, run it yourself, and pin a version. Every alternative on this page publishes source too.
- Why pick an open-source observability MCP server?
- You can audit exactly which queries and mutations an agent can issue against your telemetry, pin or patch the version, and keep credentials on infrastructure you control. Some of these connect to a hosted backend even when the client is open, so check where the data actually lives.