crewAIInc/crewAI · error · ValueError

Tavily async client is not initialized. Ensure 'tavily-pytho

Error message

Tavily async client is not initialized. Ensure 'tavily-python' is installed and API key is set.

What it means

Async twin of the sync guard: TavilyExtractorTool._arun raises this ValueError when self.async_client is None. The AsyncTavilyClient is only constructed in __init__ when tavily-python imports successfully, so a None async client means the package was missing (or init took the installer branch) at construction time.

Source

Thrown at lib/crewai-tools/src/crewai_tools/tools/tavily_extractor_tool/tavily_extractor_tool.py:166

                timeout=self.timeout,
            ),
            indent=2,
        )

    async def _arun(
        self,
        urls: list[str] | str,
    ) -> str:
        """Asynchronously extracts content from the given URL(s).

        Args:
            urls: The URL(s) to extract data from.

        Returns:
            A JSON string containing the extracted data.
        """
        if not self.async_client:
            raise ValueError(
                "Tavily async client is not initialized. Ensure 'tavily-python' is installed and API key is set."
            )

        results = await self.async_client.extract(
            urls=urls,
            extract_depth=self.extract_depth,
            include_images=self.include_images,
            timeout=self.timeout,
        )
        return json.dumps(results, indent=2)

View on GitHub (pinned to 754d7323be)

Solutions

Install tavily-python, then re-instantiate TavilyExtractorTool so async_client is built.
Confirm the api key resolves at construction (env TAVILY_API_KEY or api_key= argument).
Guard async calls with a check on tool.async_client before awaiting.

Example fix

# before
result = await tool._arun("https://example.com")  # -> ValueError

# after
if not tool.async_client:
    tool = TavilyExtractorTool()
result = await tool._arun("https://example.com")

Defensive patterns

Strategy: type-guard

Validate before calling

if not getattr(tool, "async_client", None):
    raise RuntimeError("TavilyExtractorTool has no async client; re-create the tool")

Type guard

def async_extractor_ready(tool) -> bool:
    """True when the async Tavily client is initialized."""
    return getattr(tool, "async_client", None) is not None

Try / catch

try:
    out = await tool._arun(urls)
except ValueError as e:
    if "async client is not initialized" in str(e):
        tool = TavilyExtractorTool()
        out = await tool._arun(urls)
    else:
        raise

Prevention

Verify tavily-python importability before building async crews that use Tavily tools.
Check both client and async_client attributes at tool-assembly time.
Rebuild tool objects whenever the runtime environment changes.

When it happens

Trigger: Awaiting tool._arun(urls) / tool.arun on an instance where tavily-python was unavailable during __init__, so AsyncTavilyClient was never created.

Common situations: Async CrewAI pipelines (arun-based execution) built with a stale tool instance after installing tavily-python post-hoc; mixed sync/async code paths where only the sync client was assumed.

Related errors

AI-assisted analysis of crewAIInc/crewAI@754d7323be (2026-08-15). Data as JSON: /api/errors/b5607e963e67ac8c. Report an issue: GitHub.