Configuration¶
The config file¶
It holds the selected provider and model, and any API keys you stored with K in the picker. Nothing else persists between sessions.
Permissions are enforced on every write, not only at creation: directory
0700, file 0600.
XDG_CONFIG_HOME is honoured if set, so the path follows your XDG layout.
SCELO_TUI_CONFIG overrides the whole path when you want the file somewhere
specific.
Environment¶
Choosing a model without the picker¶
| variable | does |
|---|---|
SCELO_TUI_PROVIDER |
the provider to start on |
SCELO_TUI_MODEL |
the model id. Default qwen2.5:7b-instruct-q4_K_M |
Combined with --no-intro, these make a scripted or kiosk start possible:
API keys¶
Used when nothing is stored in the config. A stored key wins over the environment.
| variable | provider |
|---|---|
ANTHROPIC_API_KEY |
Anthropic |
OPENAI_API_KEY |
OpenAI |
GEMINI_API_KEY |
|
OPENROUTER_API_KEY |
OpenRouter |
Anthropic additionally picks up an ant auth login profile.
Endpoints¶
Point a provider somewhere else, at a proxy, a gateway or a self-hosted endpoint:
| variable | default |
|---|---|
SCELO_OLLAMA_URL |
http://localhost:11434 |
ANTHROPIC_BASE_URL |
the Anthropic default |
SCELO_OPENAI_URL |
https://api.openai.com/v1 |
SCELO_GOOGLE_URL |
the Gemini default |
SCELO_OPENROUTER_URL |
https://openrouter.ai/api/v1 |
Behaviour¶
| variable | does |
|---|---|
SCELO_TUI_MOUSE=0 |
disable mouse reporting entirely for the session |
Host detection¶
Read rather than set by you, but worth knowing since setting them is how you test a host. See Whose terminal it is in.
| variable | means |
|---|---|
RSTUDIO |
an RStudio terminal |
TERM_PROGRAM=vscode, VSCODE_PID, VSCODE_CWD |
VS Code or a fork |
SCELO_IDE=1 |
a Scelo IDE terminal |
SCELO_IDE_WORKSPACE |
where that IDE's workspace is, so exports land in it |
SCELO_IDE_BIN |
the IDE binary to hand a .sce to, when it is not scelo-ide on PATH |
Command line¶
scelo # start bare
scelo data.csv # relative paths resolve against YOUR cwd
scelo ~/work/policies.csv
scelo data.csv --no-intro # skip the picker, use the saved model
What does not persist¶
The model choice and the keys survive. Nothing else does: not the loaded dataset, not the chat history, not the analysis you switched to.
Exporting the .sce at the end of a session is the workaround, and arguably the
feature. See Limits.