FAQ and troubleshooting

ModuleNotFoundError: No module named akasha

Activate the virtual environment where akasha was installed, then verify:

python -c "import akasha; print(akasha.__file__)"

Provider authentication fails

Check the environment variable name for the selected provider. Do not print the key while debugging:

import os
print(bool(os.getenv("GEMINI_API_KEY")))

Vertex AI reports DefaultCredentialsError

For API-key Vertex AI Express Mode, set only:

GEMINI_API_KEY=your_key
GOOGLE_GENAI_USE_VERTEXAI=true

Remove GOOGLE_CLOUD_PROJECT and GOOGLE_CLOUD_LOCATION. Those variables select the ADC-based full Vertex AI project/location path and take precedence over an API key.

RAG cannot find documents

Check the path from the same working directory where the program runs:

from pathlib import Path

path = Path("./docs")
print(path.resolve(), path.exists())

The answer is different each time

Model output can vary. Test the behavior you need—such as required facts or a valid event shape—instead of comparing an entire generated answer.

If an issue remains, record the Python version, akasha version, selected provider, and a redacted error message.