description: [tdd_feature_impl]: How to perform a close-loop check pattern utilizing Production DB locally for Test Driven Development (TDD)

This workflow ensures AI agents implement robust, context-safe features backed by real production data, strictly avoiding regressions.

Execution Pattern

// turbo

Sync Production DB to Local: Execute the pre-configured script to copy the production SQLite database (192.168.68.113) to your local workspace.
```
chmod +x /app/scripts/sync_prod_db.sh && bash /app/scripts/sync_prod_db.sh
```
Start Local Backend: Ensure the local backend server is running in the background. Note: Verify if it's already running using ps aux | grep uvicorn, otherwise start it.
```
cd /app/ai-hub && uvicorn app.main:app --host 0.0.0.0 --port 8000 &
```
Define the Integration Test: Based on the to-do task, write an explicit integration test locally (e.g., in /tmp/test_feature.py). Use python requests targeting http://localhost:8000 to simulate an end-to-end client API call testing the desired functionality that doesn't exist yet. The test MUST fail initially (Red step of Red-Green-Refactor).
Pause & Switch Sessions (CRITICAL): Because iterating on complex implementations causes extreme token bloating and AI context amnesia, you must PAUSE here. Explicitly provide the user with a copy-pasteable prompt template to carry into the new session:

"I have synced the production DB and staged the integration test (/tmp/...). To prevent cross-impact and memory pollution during execution, please switch to a new chat session and copy-paste this prompt:"

Hi! We are working on the Cortex AI-Hub backend (FastAPI + SQLAlchemy) and following a strict TDD close-loop workflow.

Please execute **[Task Name]** as defined in the execution plan here: @/app/docs/...

1. I have already synced the production SQLite database locally.
2. An integration test has been staged at `[Test File Path]` which hits `http://localhost:8000/...`.
3. Start the local development server in the background:
   `cd /app/ai-hub && uvicorn app.main:app --host 0.0.0.0 --port 8000 &`
4. Review the requirements from the execution plan.
5. [Insert specific files to modify here...]

You are not finished until you can successfully execute `python3 [Test File Path]` against the local server and it passes all tests automatically. Iterate on your code until it works!

Start Implementation (In New Session): Once respawned in the new session, begin architectural coding.
Validate & Loop: Every time you modify the codebase and believe you are "finished", immediately execute the integration test you defined in Step 3.
```
python3 /tmp/test_feature.py
```
If it fails, fix the code natively. You cannot declare the task completed to the user until this script runs cleanly against the local server containing the production DB dump.