Chat interface uses incorrect OpenAI API key

[caufieldjh]

Setting the OpenAI API key as stated in the README may not consistently set it in a way the app can access. If I do the following:

$ export OPENAI_API_KEY=(key_here)
$ make ont-maxo
$ cp -r stagedb/* db/
$ make app

and then use the Chat interface, I get an authentication error:

openai.error.AuthenticationError: Incorrect API key provided: sk-22ENy***************************************VnKr. You can find your API key at https://platform.openai.com/account/api-keys.
2024-02-02 12:55:41.247 Removing orphaned files...
2024-02-02 12:55:41.448 Script run finished successfully; removing expired entries from MessageCache (max_age=2)

That's not my API key...but it is one I have used in the past! It's not active anymore so it won't work here, and I'm not certain where CurateGPT is finding or why it isn't using the one I just set to OPENAI_API_KEY.

[caufieldjh]

This is the case for other operations requiring GPT access too, like Curate.

[caufieldjh]

Setting the key with runoak set-apikey -e openai (key) instead does not solve the issue.

[caufieldjh]

Ah, here's the issue: llm handles storage and retrieval of API keys in all places except the chromadb_adapter, so getting embeddings works as expected but the key this uses is pulled directly from the local environment. All other instances, like in curategpt's Extractor class, call llm.get_key() - but the key is stored with the model object returned from llm.get_model(). In practice they should all be using the same environment and have the same keys, but that may not always be true.

[caufieldjh]

llm stores its own configs, including credentials, in a user directory. These get used preferentially over environment variables. On my system they're in ~/.config/io.datasette.llm

Closing this issue as it will only happen if setting up llm independent of curategpt.