ReleaseGuard
Environment Staging only
AccountAccess and setup

Workspace overview

Release readiness

A precise operational view of what is implemented, what is configured, and what still requires verification.

Open setup guide

Recommended next action

Complete the guided setup sequence

Sign in, create a workspace, connect an authorised staging store, and see exactly which product capabilities are available now.

Get started
Connected stores
0

Live account data appears after authentication.

Test runs
0

No deterministic checkout runs are displayed yet.

Last verdict
—

A verdict requires a completed evidence-backed run.

Setup state
Open

Use the setup cockpit for live prerequisite status.

Capability status

Assurance infrastructure

Verification required

Deterministic verdict engine

Pass, fail and inconclusive states are derived from explicit assertions.

Implemented

CircleCI execution plane

Remote Chromium workflow is configured; the Bun validation job still requires a clean run.

Verify CI

Supabase control plane

Project schema and tenant-aware policies are deployed; create an account workspace to begin.

Configured

Provider-neutral investigation

AI remains disabled by default and never owns the release verdict.

Optional

Execution history

Recent runs

No data

No runs recorded

Complete account and store setup first. The test-definition editor is the next functional M5 workflow.

Review setup

Operating model

Evidence-first release flow

  1. 01

    Authenticate and create workspace

    Establish a secure account session and tenant boundary before adding store data.

  2. 02

    Connect staging

    Pair the scoped WordPress connector with an authorised non-production store.

  3. 03

    Run critical journeys

    Execute checkout flows in Chromium with deterministic assertions.

  4. 04

    Release with evidence

    Review the first divergence and ship only when required evidence is clean.

Investigation layer

Bring your own AI provider

Optional

Configure AI only after deterministic failure evidence exists. No provider is required to authenticate, connect a store, or produce a verdict.

mock
openai
gemini
openrouter
nvidia-nim
opencode
openai-compatible

Verdict isolation: provider output can explain a failure, but it cannot change the deterministic result.