React Js, Next Js Technology Upgradation (R&D)

We need to implement a robust logging service for the CaxiAIQuote application. The primary goal is to capture transaction data to debug current issues and build a dataset for fine-tuning our own AI model in the future.

Functional Requirements:

1. Non-Blocking Performance: The logging process must be asynchronous and must not slow down the main email processing loop (zero latency impact).
2. Format: Logs must be stored in JSONL (JSON Lines) format. Each line must be a valid, self-contained JSON object. This is required for compatibility with Vertex AI / OpenAI fine-tuning tools.
3. Separation of Data: * Successful Transactions: Log to `logs/training_data.jsonl`. These represent "perfect" examples (Prompt $\rightarrow$ Correct Payload). * Failed Transactions: Log to `logs/failed_prompts.jsonl`. These are for developer review to improve the system instructions.
4. Auto-Initialization: The service must automatically create the `logs` directory and the `.jsonl` files if they do not exist when the application starts.

Data to Capture (Schema):

• `timestamp`: ISO string.
• `user`: Sender email address.
• `prompt`: The original email body text.
• `status`: 'success' or 'failed'.
• `rawAiParams`: The raw JSON returned by Gemini (crucial for debugging hallucinations).
• `standardizedParams`: The validated parameters used for logic.
• `finalPayload`: The actual JSON sent to the Flight API (only for success).
• `errorMessage`: Reason for failure (only for failed).

Technical Implementation:

• Service File: Create `services/loggerService.js`.
• Method: Use Node.js `fs.createWriteStream` with the `'a'` (append) flag to keep file streams open for high performance.
• Integration: Update `index.js` (Main Worker) to call `logTransaction()` immediately after the AI processing step (Step 2).

• `services/loggerService.js`
• Updated `index.js`
• Logs folder generated in root directory (`/logs/*.jsonl`)

Need to implement a log clear functionality and archive functionality
for proper debugging