lbb_model_choice
Inspect which model each graph use runs, review trial comparisons against checked ground-truth calls, and list switches or workflow model choices.
Instructions
Model choice, read only: which model each use of a model on the graph runs, and trials that test other models against the graph's ground truth (the checked calls: a person's review, else the judge's verdict). options shows per use the model it runs now (current), LBB's model (default), a switch (switched), its checked calls, and the models a trial can test with their efforts, prices and whether the server holds their key; available=false with a reason when trials cannot run. trials lists the trials, newest first, each with its report: calls compared, both models' right share, mean score (0 to 1), cost per call and median time, the difference with its 95% interval (delta, ci_low, ci_high), outcome (too_few, better, same, worse) and qualifies (meets the bar: at least 20 calls, at most 5 points worse at the low end, no more failures, cheaper or better). trial reads one trial with its compared calls; trial_call reads one call with both answers and the ground truth. switches lists the switched uses. choices lists the model choices of the workflow uses (extract, decide, embed): without workflow the environment's row of each use and each workflow's own choice; with workflow the row in force for each of its uses (its own, else the environment's, else the default). A row has scope (environment or workflow), model, source (person or default), since, by, the model before, the use's default and options (the models it can run, with efforts, list prices and whether the server holds their key). One page in data: has_more and next_cursor page it with cursor and limit. Start, stop, switch or revert, and set or remove a choice, with lbb_model_choice_manage.
Tool call template
Replace each placeholder with a value matching the input schema. Only required parameters are included; add optional parameters as needed.
{
"id": 1,
"jsonrpc": "2.0",
"method": "tools/call",
"params": {
"arguments": {
"action": "<action>"
},
"name": "lbb_model_choice"
}
}Input Schema
| Name | Required | Description | Default |
|---|---|---|---|
| job | No | trials: only the trials of this use. | |
| use | No | choices: one use. | |
| graph | No | Graph to target; defaults to the connection's graph | |
| limit | No | trials: at most this many, newest first (default 20); choices: rows per page (default 50). | |
| action | Yes | ||
| cursor | No | choices: the next_cursor of the page before. | |
| detail | No | Response detail level. Defaults to compact. | |
| call_id | No | trial_call: a call id from the trial's calls. | |
| trial_id | No | trial / trial_call: the trial id. | |
| workflow | No | choices: a named workflow: the rows in force for it. |