crewAIInc/crewAI · warning · ValueError

URL scheme must be 'http' or 'https'

Error message

URL scheme must be 'http' or 'https'

What it means

A ValueError raised by the sync NavigateBrowserTool._run when the target URL's scheme is neither http nor https (checked via urlparse). It is then caught by the method's own `except Exception` and returned as the string "Error navigating to <url>: URL scheme must be 'http' or 'https'" rather than propagated — so callers see an error string, not an exception.

Source

Thrown at lib/crewai-tools/src/crewai_tools/aws/bedrock/browser/browser_toolkit.py:142

            return False


# Tool classes
class NavigateTool(BrowserBaseTool):
    """Tool for navigating a browser to a URL."""

    name: str = "navigate_browser"
    description: str = "Navigate a browser to the specified URL"
    args_schema: type[BaseModel] = NavigateToolInput

    def _run(self, url: str, thread_id: str = "default", **kwargs: Any) -> str:
        """Use the sync tool."""
        try:
            page = self.get_sync_page(thread_id)

            parsed_url = urlparse(url)
            if parsed_url.scheme not in ("http", "https"):
                raise ValueError("URL scheme must be 'http' or 'https'")

            # Navigate to URL
            response = page.goto(url)
            status = response.status if response else "unknown"
            return f"Navigating to {url} returned status code {status}"
        except Exception as e:
            return f"Error navigating to {url}: {e!s}"

    async def _arun(self, url: str, thread_id: str = "default", **kwargs: Any) -> str:
        """Use the async tool."""
        try:
            page = await self.get_async_page(thread_id)

            parsed_url = urlparse(url)
            if parsed_url.scheme not in ("http", "https"):
                raise ValueError("URL scheme must be 'http' or 'https'")

            # Navigate to URL

View on GitHub (pinned to 754d7323be)

Solutions

1. Normalize URLs before invoking: prepend 'https://' when the scheme is missing.
2. Reject non-http(s) input upstream (validate user/LLM URL output) — the tool intentionally blocks file:, ftp:, javascript:.
3. If you need local files, serve them over a local http server instead of file://.

Example fix

# before
result = navigate_tool._run(url="example.com")

# after
from urllib.parse import urlparse
def ensure_scheme(url: str) -> str:
    return url if urlparse(url).scheme in ("http", "https") else f"https://{url}"
result = navigate_tool._run(url=ensure_scheme("example.com"))

from urllib.parse import urlparse

def normalize_url(url: str) -> str | None:
    url = url.strip()
    if not url:
        return None
    if urlparse(url).scheme not in ("http", "https"):
        url = "https://" + url.lstrip("/")
    return url if urlparse(url).scheme in ("http", "https") else None

def is_navigable_url(url: str) -> bool:
    return urlparse(url).scheme in ("http", "https")

# The tool returns error strings instead of raising — check the return value:
result = navigate_tool._run(url=url)
if result.startswith("Error navigating"):
    url = f"https://{url}"  # or log/skip
    result = navigate_tool._run(url=url)

When it happens

Trigger: Calling the navigate_browser tool with URLs like 'ftp://example.com', 'file:///etc/passwd', 'chrome://settings', or with no scheme at all ('example.com' parses with scheme=''), or javascript: URLs from an LLM.

Common situations: An LLM agent passing a bare domain without https://; malicious or malformed URLs from user input reaching the browser tool; attempts to use file:// or internal chrome:// schemes which the toolkit deliberately blocks as a safety measure.

Related errors

• agent_id cannot be empty
• agent_id must be a string
• agent_alias_id cannot be empty
• agent_alias_id must be a string
• session_id must be a string

AI-assisted analysis of crewAIInc/crewAI@754d7323be (2026-08-15). Data as JSON: /api/errors/4dc921d6ed93e907. Report an issue: GitHub.