See everything, miss nothing
Real-time logs, performance analytics, content guardrails, and alerting. Know exactly what your functions and agent are doing.
Real-time logs
Stream logs as they happen. Filter by function, status, or time range. Search structured fields to find exactly what you need.
Live streaming
Watch logs arrive in real time as your functions execute. No polling, no refresh. Logs appear the moment they are emitted.
Function-level filtering
Isolate logs to a single function, a specific version, or a time window. Cut through the noise to find the request you care about.
Search and export
Full-text search across all log entries. Export your logs for offline analysis or long-term archival when you need a permanent record.
Performance analytics
Understand how your functions perform under real traffic. Track request volume, latency distribution, and error rates across your workspace.
Request volume and latency
See how many requests each function handles and how fast it responds. Latency percentiles (p50, p95, p99) show you the full distribution, not just the average.
Error rates and breakdowns
Track error rates per function over time. Break down failures by status code, error type, and version to pinpoint regressions quickly.
Guardrails
Content filtering applied to all dispatch paths. Protect your functions, MCP tools, A2A tasks, and pipeline steps with configurable safety policies.
Built-in checks
PII detection with regex patterns. Prompt injection defense with heuristic analysis. Toxicity scoring powered by AI. Custom regex rules for domain-specific filtering.
Three enforcement modes
Block returns a 403 and stops the request. Warn adds a header and lets it through. Log records silently for review. Choose per rule, per function.
Every dispatch path
Guardrails run on direct function calls, MCP tool invocations, A2A task submissions, and pipeline steps. One policy covers all entry points.
Alerting
Get notified before your users notice. Define alert rules with thresholds and receive notifications through the channels you already use.
Error group detection
Errors are automatically grouped by signature. When a new error group appears or an existing one spikes, you know immediately.
Multi-channel notifications
Receive alerts via email, Slack, or your Console inbox. Configure per-rule routing so critical alerts go to the right people.
Ship with confidence
Deploy with guardrails, monitor in real time, and get alerted before your users notice.
Get Started