ImportError: cannot import name 'BaseTool' from 'crewai.tools' when deploying to Cloud Run

[MCN]

Environment: macOS → Google Cloud Run (Python 3.11-slim Docker)

Issue: ImportError: cannot import name ‘BaseTool’ from ‘crewai.tools’ when deploying to Cloud Run

• Works perfectly locally

• Fails in containerized deployment

Error Location: crewai_tools/adapters/enterprise_adapter.py:7

Current Versions:

• crewai==0.175.0 (upgraded based on release notes: “fixed import issues”)

• crewai-tools==0.76.0 (latest available)

Dockerfile Strategy:

# Install local crewai source first

RUN pip install -e .

# Then install PyPI packages

RUN pip install -r requirements.txt

Question: Is the version mismatch between crewai (0.175.0) and

crewai-tools (0.76.0) causing this import conflict? What’s the recommended

approach for Cloud Run deployments? Please help

–

Request for Guidance

1. Is there a recommended approach for deploying CrewAI applications to containerized

environments like Google Cloud Run?

2. Should we expect crewai-tools to be compatible with local CrewAI source

installations?

3. Are there plans to align crewai-tools dependency versions with current CrewAI

releases?

4. Is there a workaround for the BaseTool export discrepancy between local source and

PyPI packages?

[carmelo_iaria]

Hi @MCN, your local environment tolerates the mixed installation, but the containerized Cloud Run environment enforces stricter import resolution.

Standardizing on PyPI packages solves this consistently.

1. Recommended approach for Cloud Run deployments?

Use PyPI packages only - avoid mixing local source installations (pip install -e .) with PyPI packages. The pip install 'crewai[tools]' syntax is the official recommendation for production deployments.

2. Should crewai-tools be compatible with local source installations?

No, this is not officially supported. The crewai-tools package expects BaseTool to be available from the installed crewai package’s public API. Local editable installs can cause import resolution conflicts in containerized environments.

3. Are there plans to align versions?

Based on the release notes (v0.175.0), the team is aware of import issues and has been fixing them. However, the version numbering mismatch (crewai 0.175.0 vs crewai-tools 0.76.0) suggests these are separate release cycles. Use crewai[tools] to ensure compatibility.

4. Workaround for BaseTool export discrepancy?

The fix: Ensure BaseTool is imported from crewai.tools:

python

from crewai.tools import BaseTool

If crewai_tools is trying to import it, make sure you’re using compatible versions installed from PyPI together.

Also, make sure you test Docker image locally before Cloud Run deployment

docker build -t crewai-test .

# Run
docker run --rm crewai-test

# If you need to inspect
docker run -it --rm crewai-test /bin/bash

let me know if this helps!