Operations dashboard

Watch exchange rate, revocation cascades, federation peers, and MCP denials in one place — read-only, off the hot path. The operations dashboard is the human NOC for a fabric unit. If it goes away, tokens still exchange and revocations still propagate; you lose visibility, not security.

Why it’s worth your time

• Live incident picture — SSE event feed plus Prometheus metrics on one screen.
• Demo-ready — show delegation depth, federation topology, and per-tool MCP outcomes without scripting curls.
• Zero PEP risk — the dashboard only reads /metrics, /v1/events, and health endpoints.

What you get

Tab	What you see
Fabric Pulse	Exchange rate, latency, active agents, live SSE
Delegation	Depth histogram, blast-radius denials
MCP Security	Per-tool verification and policy denials
Federation	Peer health, revocation relay, CAEP cascade
Soul	Manifest timeline and recovery readiness
Trust Tree	Trust domain hierarchy

Screenshots: starfly.dev/play — live captures from the lab dashboard

How it connects

Dashboard  →  GET /metrics, /v1/events, /v1/sys/health, /v1/mcp/tools, /v1/sys/trust-domains
                (all read-only, proxied server-side)

No writes to the PEP. Revocation and federation actions still go through signals API or operator tooling.

When to use what

Goal	Use
Watch a demo or incident	Dashboard Pulse + SSE
Scriptable health check	curl $STARFLY_URL/v1/sys/health
Agent integration	Token exchange, UTC

Code in this repo

Path	Status
dashboard/	Preview — Next.js app export pending

Deploy with Helm (dashboard.enabled=true) when the app slice is published here.

Related

• UTC · Starfly Graph — other async integrator surfaces
• Getting started — stand up a fabric unit first
• Documentation voice