diff --git a/Data/Engine/CURRENT_STAGE.md b/Data/Engine/CURRENT_STAGE.md
index d478102..30cf14b 100644
--- a/Data/Engine/CURRENT_STAGE.md
+++ b/Data/Engine/CURRENT_STAGE.md
@@ -59,9 +59,11 @@
 - 11.2 Provide repository/service hooks for fetching artifacts or repo heads; add resilience logging.
 - 11.3 Commit after integration tests (or mocked unit tests) confirm API workflows.
 
-12. Final parity verification
-- 12.1 Stand up Engine end-to-end in a staging environment, exercising enrollment, token refresh, agent connections, and jobs.
-- 12.2 Document any divergences and address them with follow-up commits.
-- 12.3 Once satisfied, coordinate cut-over steps (switch entrypoint, deprecate legacy server) as a future initiative.
-- Documentation and test coverage for this phase now live in `Data/Engine/README.md` and `Data/Engine/tests/` to guide the
-  remaining staging work.
+[IN PROGRESS] 12. Final parity verification
+- 12.1 Follow the staging playbook in `Data/Engine/STAGING_GUIDE.md` to stand up the Engine end-to-end and exercise enrollment,
+  token refresh, agent connections, GitHub integration, and scheduler flows.
+- 12.2 Record any divergences in the staging guide’s table and address them with follow-up commits before cut-over.
+- 12.3 Once parity is confirmed, coordinate entrypoint switching (point deployment at `Data/Engine/bootstrapper.py`) and plan
+  the legacy server deprecation.
+- Supporting documentation and unit tests live in `Data/Engine/README.md`, `Data/Engine/STAGING_GUIDE.md`, and
+  `Data/Engine/tests/` to guide the remaining staging work.
diff --git a/Data/Engine/README.md b/Data/Engine/README.md
index 18de81a..1b9d433 100644
--- a/Data/Engine/README.md
+++ b/Data/Engine/README.md
@@ -130,13 +130,16 @@ The service container now wires `github_service`, giving other interfaces and ba
 ## Final parity checklist
 
 Step 12 tracks the final integration work required before switching over to the
-Engine entrypoint:
+Engine entrypoint. Use the detailed playbook in
+[`Data/Engine/STAGING_GUIDE.md`](./STAGING_GUIDE.md) to coordinate each
+staging run:
 
 1. Stand up the Engine in a staging environment and exercise enrollment, token
    refresh, scheduler operations, and the agent real-time channel side-by-side
    with the legacy server.
-2. Capture any behavioural differences uncovered during staging and file them
-   for follow-up fixes before the cut-over.
+2. Capture any behavioural differences uncovered during staging using the
+   divergence table in the staging guide and file them for follow-up fixes
+   before the cut-over.
 3. When satisfied with parity, coordinate the entrypoint swap (point production
    tooling at `Data/Engine/bootstrapper.py`) and plan the deprecation of
    `Data/Server`.
diff --git a/Data/Engine/STAGING_GUIDE.md b/Data/Engine/STAGING_GUIDE.md
new file mode 100644
index 0000000..60ae3b5
--- /dev/null
+++ b/Data/Engine/STAGING_GUIDE.md
@@ -0,0 +1,116 @@
+# Engine Staging & Parity Guide
+
+This guide supports Step 12 of the migration plan by walking operators through
+standing up the Engine alongside the legacy server, validating core workflows,
+and documenting any behavioural gaps before switching the production entrypoint
+to `Data/Engine/bootstrapper.py`.
+
+## 1. Prerequisites
+
+- Python 3.11 or later available on the host.
+- A clone of the Borealis repository with the Engine tree checked out.
+- Access to the legacy runtime assets (certificates, TLS bundle, etc.).
+- Optional: a staging agent install for end-to-end WebSocket validation.
+
+Ensure the SQLite database lives at `<project_root>/database.db` and that the
+Engine migrations have already run (they execute automatically when the
+`BOREALIS_ENGINE_AUTO_MIGRATE` environment variable is left at its default
+`true`).
+
+## 2. Launching the Engine in staging mode
+
+1. Open a terminal at the project root.
+2. Set any environment overrides required for the test scenario (for example,
+   `BOREALIS_DEBUG=true` to surface verbose logging, or
+   `BOREALIS_CORS_ALLOWED_ORIGINS=https://localhost:3000` when pairing with the
+   React UI).
+3. Run the Engine entrypoint:
+
+   ```bash
+   python Data/Engine/bootstrapper.py
+   ```
+
+4. Verify `Logs/Server/engine.log` is created and that the startup entries are
+   timestamped `<timestamp>-engine-<message>`.
+
+Keep the legacy server running in a separate process if comparative testing is
+required; they do not share global state.
+
+## 3. Feature validation checklist
+
+Work through the following areas and tick each box once verified. Capture any
+issues in the log table in §4.
+
+### Authentication and tokens
+
+- [ ] `POST /api/agent/token/refresh` returns a new access token when supplied a
+      valid refresh token + DPoP proof.
+- [ ] Invalid DPoP proofs or revoked refresh tokens yield the expected HTTP 401
+      responses and structured error payloads.
+- [ ] Device last-seen metadata updates inside the database after a successful
+      refresh.
+
+### Enrollment
+
+- [ ] `POST /api/agent/enroll/request` produces an enrollment ticket with the
+      correct expiration and retry counters.
+- [ ] `POST /api/agent/enroll/poll` transitions an approved device into an
+      authenticated state and returns the TLS bundle.
+- [ ] Audit logging for approvals lands in `Logs/Server/engine.log`.
+
+### Job management
+
+- [ ] `POST /api/jobs` (or UI equivalent) creates a scheduled job and returns a
+      manifest identifier.
+- [ ] `GET /api/jobs/<id>` surfaces the stored manifest with normalized
+      schedules and environment variables.
+- [ ] Job lifecycle events arrive over the `job_management` Socket.IO namespace
+      when a job transitions between `pending`, `running`, and `completed`.
+
+### Real-time agents
+
+- [ ] Agents connecting to the `agents` namespace appear in the realtime roster
+      with accurate hostname, username, and fingerprint details.
+- [ ] Screenshot broadcasts relay from agents to the UI without residual cache
+      bleed-through after disconnects.
+- [ ] Macro execution responses round-trip through Socket.IO and reach the
+      initiating client.
+
+### GitHub integration
+
+- [ ] `GET /api/repo/current_hash` reflects the latest branch head and caches
+      repeated calls.
+- [ ] `POST /api/github/token` persists a new token and survives Engine restarts
+      (confirm via database inspection).
+- [ ] The background refresher logs rate-limit warnings instead of raising
+      uncaught exceptions when the GitHub API throttles requests.
+
+## 4. Recording divergences
+
+Use the table below to document behavioural differences or bugs uncovered during
+staging. This artifact should accompany the staging run summary so follow-up
+fixes can be triaged quickly.
+
+| Area | Legacy Behaviour | Engine Behaviour | Notes / Links |
+| --- | --- | --- | --- |
+| Authentication |  |  |  |
+| Enrollment |  |  |  |
+| Scheduler |  |  |  |
+| Realtime |  |  |  |
+| GitHub |  |  |  |
+| Other |  |  |  |
+
+## 5. Cut-over readiness
+
+Once every checklist item passes and no critical divergences remain:
+
+1. Update `Data/Engine/CURRENT_STAGE.md` with the completion date for Step 12.
+2. Coordinate with the operator to switch deployment scripts to
+   `Data/Engine/bootstrapper.py`.
+3. Plan a rollback strategy (typically re-launching the legacy server) should
+   issues appear immediately after the cut-over.
+4. Archive the filled divergence table alongside Engine logs for historical
+   traceability.
+
+Document the results in project tracking tools before moving on to deprecating
+`Data/Server`.