Analytics & Observability

Fp Switchboard tracks every tool call, providing usage analytics, service health monitoring, and execution observability for both users and admins.

Usage Metering

Every MCP tool call is metered and aggregated into daily summaries per user and service.

How It Works

Tool call executes and is logged in the audit trail
Daily CRON job aggregates logs into usage_daily_summaries table
User and admin dashboards query the rollup data
CSV/JSON export available for all time periods

Data Tracked

Metric	Granularity	Retention
Total tool calls	Per user, per service, per day	90+ days
Success/failure rate	Per service, per day	90+ days
Error breakdown	Per tool, per error type	90+ days
Response latency	Per service	90+ days

User Analytics

Available at Dashboard > Analytics for all users.

Dashboard Panels

Daily Call Volume

Bar chart showing tool calls per day with trend line

Service Breakdown

Which services you use most, with percentage bars

Top Tools

Your most-used tools ranked by call count

Error Rate Trend

Error rate over time with anomaly highlighting

Time Range Selector

All analytics views support:

Today — Current day activity
7 days — Past week
30 days — Past month
Custom — Pick specific start/end dates

Export

Click the export button on any analytics view to download:

CSV — Spreadsheet-compatible format
JSON — Structured data for programmatic use

Admin Analytics

Available at Admin > Analytics for org admins and platform admins.

Admin-Only Views

Panel	Description
My/All Toggle	Switch between personal and organization-wide analytics
User Activity Ranking	Users ranked by total calls with activity bars
Service Adoption	Which services are used across the organization
Error Hotspots	Services and tools with highest error rates

Tool Observability

Every tool execution is captured with DLP-sanitized detail.

Execution Detail View

Click any activity item to see:

Status — Success, failure, or DLP blocked
Timing — Request duration in milliseconds
Error Details — Full error message with actionable fix suggestions
AI Intent — What the AI was trying to accomplish
Request Args — DLP-sanitized input parameters
Response Preview — Truncated response (500 chars)

Error Drill-Down

Failed calls show categorized errors with fix suggestions:

Error Type	Suggestion
Auth failure	Re-connect the service, refresh OAuth token
Permission denied	Check scopes, request additional permissions
Rate limited	Wait and retry, reduce request frequency
Timeout	Retry the call, check service status
Not found	Verify resource ID, check access permissions

Real-Time Feed

Enable the Live button for real-time monitoring:

5-second polling interval
Incremental updates (only new events)
Auto-scrolls to latest events

Service Health

The health panel shows 24-hour status for each connected service:

Color	Meaning
Green	Above 95% success rate
Yellow	80–95% success rate
Red	Below 80% success rate

Each tile displays:

Total calls in 24 hours
Success rate percentage
Average response latency

Usage APIs

User Usage API

GET /dashboard/api/usage?period=7d
Authorization: (Cloudflare Access session)

Returns daily breakdown with service-level detail.

Admin Usage API

GET /admin/api/usage?period=30d
Authorization: (Cloudflare Access session, admin role)

Returns org-wide rollup with user ranking and service breakdown.

Period Options

Parameter	Description
`today`	Current day
`7d`	Past 7 days
`30d`	Past 30 days
`custom`	Use `start` and `end` query params (ISO 8601)