Obtain HttpOnly=True Session Cookies from Services Logged in via Browser-Use
2025-05-21 12:10 (2 days ago)
ytyng
Here is how you can obtain session cookies issued when logging into a web service using browser-use.
Session cookies are typically issued with httponly=true, so even if you request browser-use to "retrieve session cookies," they won't be returned. This is because they cannot be accessed via JavaScript.
If you want to obtain them, you can prepare a browser_use.Browser instance in advance and inject it into the browser-use Agent, allowing you to handle the Playwright browser instance during processing.
This approach is effective not only for session cookies but also when dealing with finer resources inside the browser.
from browser_use import Agent, Browser
from langchain_openai import ChatOpenAI
DEFAULT_LLM_MODEL = "gpt-4.1-mini"
def get_session_cookies():
browser = Browser()
current_cookies = []
async def handle_step(browser_state, agent_output, step_number):
nonlocal browser, current_cookies
try:
# Handle the Playwright instance
context = browser.playwright_browser.contexts[0]
current_cookies = list(await context.cookies())
except Exception as e:
print(
'An error occurred while getting cookies: '
f'{e.__class__.__name__}: {e}'
)
_login_prompt = 'Please log into our web service ◯◯.\n\n...'
async def run_agent():
agent = Agent(
task=_login_prompt,
browser=browser,
llm=ChatOpenAI(model=DEFAULT_LLM_MODEL),
register_new_step_callback=handle_step,
)
await agent.run(max_steps=5)
asyncio.run(run_agent())
# Login complete. Cookies should be obtained here
print(current_cookies)
The author runs the application development company Cyberneura.
We look forward to discussing your development needs.
We look forward to discussing your development needs.