What it does

Exposes Google Cloud Observability APIs—Cloud Logging, Cloud Metrics, and Cloud Trace—as MCP tools that Claude can call directly. The server bridges 13 tools for querying logs across projects, analyzing time-series metrics, and inspecting distributed traces without switching context or manually constructing gcloud commands.

Who it's for

GCP operators, SRE engineers, and backend developers investigating production incidents, debugging performance bottlenecks, or monitoring cloud workload health. Useful for teams already using gcloud CLI and wanting natural-language access to observability data from Claude.

Common use cases

Query Cloud Logging to search error patterns or audit logs across multiple GCP projects.
Analyze Cloud Metrics time series to correlate performance spikes with application changes.
Inspect Cloud Trace spans for request latency breakdown and service dependency analysis.
Aggregate observability signals (logs, metrics, traces) during incident investigation.

Setup pitfalls

Requires both the Node.js package @google-cloud/observability-mcp and the gcloud CLI installed and authenticated; the package alone is insufficient.
Node.js 20 or higher required; earlier versions will not work.
The GCP service account or user credentials must have authorization to access Cloud Logging, Metrics, and Trace APIs.
CI is currently failing as of the latest commit; verify compatibility with your Claude client before upgrading.

Tool name

Description

Destructive?

list_log_entries

✓ no

list_log_names

✓ no

list_buckets

✓ no

list_views

✓ no

list_sinks

✓ no

list_log_scopes		✓ no
list_metric_descriptors		✓ no
list_time_series		✓ no
list_alert_policies		✓ no
list_alerts		✓ no
list_traces		✓ no
get_trace		✓ no
list_group_stats		✓ no

@google-cloud/observability-mcp

What it does

Who it's for

Common use cases

Setup pitfalls