An authenticated control centre answering what the buyer actually asks about the software run for them: what is live and is it up, which machines it runs on, how the pieces fit together, whether tests prove it works, whether preview matches live, and what was billed for what. Six routed, deep-linkable views signed in against their own directory, with the drift list (what two records disagree about) as the section customers value most.
What ports as-is
- The route table and legacy aliases
- App-card and group-disclosure anatomy
- Shared page chrome
- The apps cockpit with live per-slot health probes
- A lazily loaded deployment graph
- The quality view with report producer and gap derivation
- The configuration view with preview-versus-live diff classification
- The billing journey with its gateway and store
- Five data readers with no backend of its own
- An nginx container with runtime configuration injected at start so one image runs in any slot
- Rule zero (every figure transcribed from a named artifact, absent facts rendered as designed empty states with their reason, secrets shown as names and locations only)
- A named scaffold workflow with six reference leaves that lands a building
- On-brand portal in about an hour of agent time
What we build for you
- Capture the brand from their running site
- Point the readers at that estate's real artifacts (deployment manifest, compose files, edge config, time ledger, test output, commit history)
- Take the roster and card labels verbatim from the owner
- Wire sign-in to their directory reusing an existing app registration where one exists
- Supply the rate and hosting decisions that are not on disk
- Write the provenance document
