Skip to main content
LlamaIndex connects LLMs to your data through loaders, indexes, and retrievers. Each of those typically needs a credential (LLM API key, data-source token). Run the script under authsome’s proxy and the loaders just work.

​
Run the whole script under the proxy

authsome login openai
authsome login github
authsome run -- python my_index.py
LlamaIndex’s HTTP clients see OPENAI_API_KEY=authsome-proxy-managed and similar placeholders. The proxy injects the real value for outbound requests to matched provider hosts. No code changes needed in your script.

​
Data-loader credentials

LlamaIndex’s data readers cover dozens of services (Notion, Confluence, Slack, Google Drive, etc.). For each one:
  • Check authsome list for the matching provider name.
  • If bundled, log in: authsome login <provider>.
  • Run your script under authsome run --. The reader picks the credential up from its standard env var.
For services not bundled, see Custom providers.

​
Export when the proxy is not enough

eval "$(authsome export openai --format env)"
eval "$(authsome export github --format env)"
python my_index.py
LlamaIndex clients read the exported env vars as usual.

​
Multi-account

eval "$(authsome export github --connection personal --format env)"
eval "$(authsome export github --connection work --format env)"
See Multiple connections per provider.

​
What’s next

Python library

the CLI export path or the daemon HTTP API.

Custom providers

Add a data-source provider authsome doesn’t ship.