Internal Dashboard — Overview

The Internal Dashboard (Surface 02) is the infrastructure monitoring surface for the GOVERN platform. It is not a customer-facing product — it is how the GOVERN civilization monitors, understands, and improves itself.

The Concept: A Civilization Monitoring Itself

GOVERN is a civilization that builds monitoring civilizations for customers. The Internal Dashboard is the mechanism through which the parent civilization maintains self-awareness.

Every build event, every archetype skill evolution, every convergence score, every token spent — all of it flows through the Internal Dashboard. Engineers and operations teams use it to understand what the civilization is doing, whether it is healthy, and where it needs attention.

This is not a vanity dashboard. Every panel is wired to real data from production systems.

What the Internal Dashboard Shows

Real-time feeds

Panel	Data source	Update frequency
Pipeline Monitor	`monitoring_pipeline` Supabase table	Real-time (Supabase Realtime)
Build Event Stream	`build_events` table	Real-time
Active Agents	`agent_task_assignments` DO	5-second poll
Recent Deploys	Cloudflare Pages API	60-second poll

Aggregated views

Panel	Data source	Aggregation period
Convergence Scores	`monitoring_rollup` table	Daily, weekly, monthly
Skill Evolution Graph	`archetype_skills` table	Cumulative
Cost Trends	`token_usage` table	Hourly, daily, monthly
Deploy Health	`deploy_history` table	Rolling 30 days

Architecture

The Internal Dashboard is a React SPA (packages/govern-dashboard) that reads from two data sources:

GOVERN API (/api/monitoring/*, /api/build-events, /api/assess/*) — operational data
Supabase Realtime — live streaming for pipeline events and build events

Data flow

[Cloudflare Worker — API Gateway]
  ↓ build events, monitoring emit
[Monitoring Pipeline]
  EMIT → ACCUMULATE → ROLLUP → INSCRIBE → RECALL → BUILD
  ↓ persisted to Supabase
[Internal Dashboard]
  → reads via API + Realtime subscriptions
  → displays in React panels

The Six Pipeline Stages

The monitoring pipeline is the backbone of the Internal Dashboard. All data flows through these stages:

EMIT     → Event arrives (build event, monitoring event, assess result)
ACCUMULATE → Events collected in memory window (60-second batches)
ROLLUP   → Aggregated into summary metrics (hourly, daily, weekly)
INSCRIBE → Written to Supabase for persistence and querying
RECALL   → Historical data retrieved for dashboard views
BUILD    → Metrics consumed by convergence engine and skill evolution

Full documentation: Pipeline Monitoring

Dashboard Layout

The Internal Dashboard is organized into five sections:

1. Top bar — Current status indicators: API health, pipeline status, deploy health, active incidents

2. Pipeline Monitor — Live view of events flowing through the pipeline. Each stage shows its current throughput, error rate, and queue depth.

3. Build Activity — Recent build events, archetype activity, skill evolution triggers. The primary view for understanding what the civilization is building.

4. Quality & Convergence — V(Q) scores over time, gate pass rates, test failure trends. For engineers tracking release quality.

5. Infrastructure — Cost governor, deploy watchdog, Worker health, Supabase connection pool status.

Access Control

The Internal Dashboard is restricted to Archetypal AI team members. It is not accessible to customers.

Role	Access
Engineer	Full read access
Operations	Full read + incident management
Admin	Full read + write (can trigger rollbacks, reset cost alerts)

Authentication uses the same Bearer token as the GOVERN API. A separate INTERNAL_DASHBOARD_SECRET must be set in the Worker environment.

Relationship to Customer Dashboard

The Internal Dashboard and Customer Dashboard share some data models but serve different purposes:

	Customer Dashboard	Internal Dashboard
Audience	Customer’s ops team	Archetypal AI team
Data scope	One customer org	All orgs + infrastructure
Primary purpose	AI system governance	Civilization self-monitoring
Real-time feeds	Assessment status	Pipeline throughput
Cost visibility	Customer’s AI spend	Platform infrastructure spend

The Internal Dashboard can see everything. The Customer Dashboard sees only the customer’s own data.

Monitoring Principles

The Internal Dashboard follows three operational principles:

1. Wire first, decorate later Every panel shows real data or shows an error. No placeholder numbers, no mock data in production, no “coming soon” widgets. If data is not yet wired, the panel is not shipped.

2. Anomalies are surfaces, not alarms The dashboard surfaces anomalies for human judgment. It does not auto-respond. A cost spike shows up in the dashboard; a human decides whether to act. The exception is hard budget limits, which trigger automatic alerts.

3. The civilization’s health is a single number The civilizationHealthScore (displayed in the top bar) is a composite metric across pipeline throughput, convergence scores, deploy health, and cost governor status. When this number is green, the civilization is healthy.