Cloudflare Workers Observability

Debug and monitor Cloudflare Workers using logs and analytics from the Observability MCP server.

Available Tools

Tool	Purpose
query_worker_observability	Query logs and metrics from Workers
observability_keys	Discover available data fields in logs
observability_values	Find available values for specific fields

Query Workflow

1. Discover Available Fields

Use observability_keys to find what data is available:

• Metadata fields (timestamps, status codes)
• Worker-specific fields (script name, route)
• Custom logged fields from console.log

2. Explore Field Values

Use observability_values to find valid values for filtering:

• Status codes present in logs
• Script names deployed
• Custom field values

3. Query Logs and Metrics

Use query_worker_observability to:

• List recent events/invocations
• Calculate metrics (error rates, latency)
• Find specific invocations by criteria

Common Queries

Goal	Approach
Recent errors	Query for events with error status
Latency analysis	Query for execution time metrics
Traffic patterns	Query for invocation counts over time
Specific request	Query by request ID or timestamp
Script comparison	Query metrics grouped by script name

Debugging Workflow

1. Identify the problem

   • Query recent errors with query_worker_observability

2. Find patterns

   • Use observability_keys to discover relevant fields
   • Use observability_values to see error types

3. Narrow down

   • Add filters for specific routes, times, or status codes

4. Analyze specific invocations

   • Query for detailed logs of problematic requests

Post-Deployment Monitoring

After deploying, check for issues:

1. Query for errors in the last 5-10 minutes
2. Compare error rates before/after deployment
3. Check latency metrics for performance regression

Tips

• Start broad, then add filters to narrow results
• Use observability_keys when unsure what fields exist
• Custom console.log output appears in queryable fields
• Combine with builds tools to correlate issues with deployments