unclecode/crawl4ai · error · RuntimeError

Browser is not available. It may have been closed, crashed,

Error message

Browser is not available. It may have been closed, crashed, or not yet started. Ensure the browser is running before creating new contexts.

What it means

The sibling guard in create_browser_context(): self.browser is None and the manager was NOT launched persistently, meaning the Playwright Browser handle is gone — it was never started, was closed, or crashed. Any subsequent context/session creation hits this RuntimeError.

Source

Thrown at crawl4ai/browser_manager.py:1268

            for script in self.config.init_scripts:
                await context.add_init_script(script)

    async def create_browser_context(self, crawlerRunConfig: CrawlerRunConfig = None):
        """
        Creates and returns a new browser context with configured settings.
        Applies text-only mode settings if text_mode is enabled in config.

        Returns:
            Context: Browser context object with the specified configurations
        """
        if self.browser is None:
            if self._launched_persistent:
                raise RuntimeError(
                    "Cannot create new browser contexts when using "
                    "use_persistent_context=True. Persistent context uses a "
                    "single shared context."
                )
            raise RuntimeError(
                "Browser is not available. It may have been closed, crashed, "
                "or not yet started. Ensure the browser is running before "
                "creating new contexts."
            )
        # Base settings
        user_agent = self.config.headers.get("User-Agent", self.config.user_agent) 
        viewport_settings = {
            "width": self.config.viewport_width,
            "height": self.config.viewport_height,
        }
        proxy_settings = {"server": self.config.proxy} if self.config.proxy else None

        # CSS extensions (blocked separately via avoid_css flag)
        css_extensions = ["css", "less", "scss", "sass"]

        # Static resource extensions (blocked when text_mode is enabled)
        static_extensions = [
            # Images

View on GitHub (pinned to 7e80152142)

Solutions

Use the crawler inside its lifecycle: async with AsyncWebCrawler() as crawler: await crawler.arun(...) — it auto-starts.
If you manage lifecycle manually, await crawler.start() before use and don't call methods after close().
Create a fresh AsyncWebCrawler instance per batch instead of reusing a closed one.
Check earlier logs for the browser crash/stop reason (startup failure, OOM) and fix that first.

Example fix

# before
crawler = AsyncWebCrawler()
await crawler.start()
await crawler.close()
results = await crawler.arun(url)  # RuntimeError: browser not available

# after
async with AsyncWebCrawler() as crawler:
    results = await crawler.arun(url)

Defensive patterns

Strategy: validation

Validate before calling

if crawler.crawler_strategy.browser is None and not crawler.ready:
    await crawler.start()  # or refuse to proceed

Type guard

async def is_crawler_ready(crawler) -> bool:
    return bool(getattr(crawler, 'ready', False)) or (
        getattr(crawler.crawler_strategy, 'browser', None) is not None
    )

Try / catch

try:
    results = await crawler.arun(url)
except RuntimeError as e:
    if 'Browser is not available' in str(e):
        crawler = AsyncWebCrawler(config=browser_config)  # fresh instance
        async with crawler:
            results = await crawler.arun(url)
    else:
        raise

Prevention

Always use 'async with AsyncWebCrawler() as crawler' scoping.
Never call crawler methods after close().
Create a new instance per batch rather than resurrecting closed ones.

When it happens

Trigger: Calling crawler.arun() after __aexit__ closed the browser; reusing a crawler instance after its browser crashed (see 'Failed to start browser'); calling internal APIs before awaiting crawler.start(); using a manager whose cleanup() already ran.

Common situations: Reusing an AsyncWebCrawler outside its async with block; forgetting await crawler.start() when calling low-level strategy methods; background tasks racing with shutdown; browser process killed by OOM.

Related errors

AI-assisted analysis of unclecode/crawl4ai@7e80152142 (2026-08-14). Data as JSON: /api/errors/0251bff83bb12388. Report an issue: GitHub.