How AI-Powered DeepSeek Can Auto‑Heal Your Playwright Tests

1. Introduction

Automation testing often suffers from fragility: scripts may fail intermittently, break after UI updates, or error due to minor element changes. This guide shows how to leverage the DeepSeek programming model with Playwright to predict failure reasons, automatically adjust locators, and generate repair code.

2. Practical Example

2.1 Install dependencies

pip install openai playwright python-dotenv
playwright install

Add your DeepSeek API key to a .env file:

DEESEEK_API_KEY=your_deepseek_key_here

2.2 Write self‑healing module

# heal/deepseek_client.py
import os
import openai
from dotenv import load_dotenv

load_dotenv()
openai.api_key = os.getenv("DEESEEK_API_KEY")

def suggest_fix(code_context: str, error_msg: str) -> str:
    prompt = f"""You are an automation testing expert. Here's the playwright script context and error:

---code---
{code_context}
---error---
{error_msg}

Please analyze the root cause and suggest the corrected code snippet."""
    response = openai.ChatCompletion.create(
        model="deepseek-coder",
        messages=[{"role": "user", "content": prompt}],
        temperature=0.3,
    )
    return response["choices"][0]["message"]["content"]

2.3 Trigger failure and call healing

# tests/test_checkout_healing.py
from playwright.sync_api import sync_playwright
from heal.deepseek_client import suggest_fix

def test_with_healing():
    try:
        with sync_playwright() as p:
            browser = p.chromium.launch(headless=False)
            page = browser.new_page()
            page.goto("https://www.saucedemo.com")
            page.fill("#user-name", "standard_user")
            page.fill("#password", "secret_sauce")
            # intentional wrong selector to simulate failure
            page.click("#login-btn")
    except Exception as e:
        print("[❌ Test failed] Captured exception, invoking DeepSeek for fix...")
        code_context = '''
page.fill("#user-name", "standard_user")
page.fill("#password", "secret_sauce")
page.click("#login-btn")
'''
        fix = suggest_fix(code_context, str(e))
        print("✅ DeepSeek suggestion:", fix)

Running the script outputs a suggestion to replace #login-btn with #login-button, fixing the selector error.

3. Conclusion

The AI‑driven approach can analyze failure contexts, propose corrective code, and even generate patches, moving toward an “AI test engineer” capable of writing, repairing, and reporting tests automatically.