Tool Reference
Complete reference for every tool exposed by the xplainable MCP server. Tools are grouped by category.
Select a tool, review the sample arguments, and invoke it to see a simulated response.
Discovery & Session
list_user_teams
List all teams the authenticated user belongs to.
No parameters.
select_team
Interactively select the active team for this session.
No parameters.
set_active_team
Set the active team for this session.
team_idstrRequired
The team ID to switch to (from list_user_teams).
Models
models_apply_relationships
Re-apply the training dataset's current feature relationships to an existing (v2 / XGM) model version.
version_idstrRequired
Model version to reconfigure.
dataset_idstrdefault: None
Dataset whose declaration to apply; defaults to the dataset the version was trained on.
models_get_feature_info
Get feature information including types, health metrics, and distributions.
version_idstrRequired
ID of the model version.
models_get_model
Get detailed information about a model.
model_idstrRequired
ID of the model
models_get_model_evaluation
Get detailed evaluation metrics for a model partition.
partition_idstrRequired
ID of the model partition.
models_get_model_profile
Get the model profile showing feature contributions and decision boundaries.
version_idstrRequired
ID of the model version.
models_link_preprocessor
Link a model version to a preprocessor version.
model_version_idstrRequired
The model version ID
preprocessor_version_idstrRequired
The preprocessor version ID
models_list_model_versions
List all versions of a model.
model_idstrRequired
ID of the model
models_list_team_models
List all models for the current team (based on API key).
No parameters.
models_refit_features
Refit chosen features of a trained (v2 / XGM) model into a new version.
version_idstrRequired
The model version to refit (must be v2 / XGM).
dataset_idstrRequired
The dataset it was trained on (raw; the linked preprocessor is applied server-side).
target_columnstrRequired
The target column, as at training.
feature_paramsDictRequired
{feature: {knob: value}}, e.g. {"Tenure": {"l2": 50, "monotonic": "decreasing"}}.
drop_columnsList[str]default: None
Columns dropped at training (same list).
test_sizefloatdefault: 0.2
Test split fraction, as at training.
seedintdefault: 42
Split seed, as at training (default 42), so metrics are comparable across versions.
models_train_model
Train a new model server-side from a platform dataset.
dataset_idstrRequired
ID of the dataset on the platform.
target_columnstrRequired
Name of the column to predict.
model_namestrRequired
Name for the new model.
model_descriptionstrdefault: ''
Optional description of the model's purpose.
model_typestrdefault: 'classification'
"classification" or "regression".
feature_columnsList[str]default: None
Explicit feature whitelist. Defaults to all columns except the target and drop_columns.
drop_columnsList[str]default: None
Columns to exclude (ids, leakage, etc.).
preprocessor_version_idstrdefault: None
Optional fitted preprocessor version to apply before training.
monotonic_featuresDictdefault: None
Monotonic constraints, as a map of numeric feature name to direction: "increasing" (higher value never lowers the prediction) or "decreasing". e.g. {"Monthly Charges": "increasing", "Tenure": "decreasing"}. Applied to the whole effect, interactions included. Merged over the dataset's declared monotonic relationships (explicit wins). Constraints on non-numeric features are ignored.
test_sizefloatdefault: 0.2
Test split fraction (0 < test_size < 1).
seedintdefault: 42
Random seed for the train/test split. The dataset's declared feature relationships (datasets set_relationships: derived columns, implications, monotonic) are copied into the version automatically, so the optimiser enforces them; declare them before training.
Auto-Train
autotrain_summarize_by_dataset_id
Summarize a dataset that's already on the xplainable platform.
dataset_idstrRequired
ID of the dataset on the platform
team_idstrdefault: None
Team ID (uses session team_id if not provided)
Datasets
datasets_get_dataset_info
Get information about a specific dataset.
dataset_idstrRequired
ID of the dataset
datasets_get_relationships
Read the dataset's declared feature relationships.
datasets_infer_relationships
Propose feature relationships from the data, with evidence.
dataset_idstrRequired
Dataset to scan.
target_columnstrdefault: None
Target column, for monotonic hints (optional).
datasets_list_team_datasets
List all datasets for a team.
team_idstrdefault: None
Optional team ID (uses session team_id if not provided)
datasets_preview_dataset_json
Preview a dataset as JSON records.
dataset_idstrRequired
ID of the dataset
rowsintdefault: 10
Number of rows to return (1-1000)
offsetintdefault: 0
Rows to skip before the window (ignored when sampling)
samplebooldefault: False
Return a reproducible random sample instead of a head window
datasets_set_relationships
Declare (replace) the dataset's feature relationships.
dataset_idstrRequired
Dataset to declare on.
derivedDictdefault: None
``{column: expression}``, e.g. ``{"EstimatedLifetimeCharges": "tenure * MonthlyCharges"}``. The optimiser never moves a derived column and recomputes it from its parents for every candidate.
impliesList[Dict]default: None
``[{"when": {feature: [values]}, "then": {feature: [allowed values]}}]``. Compiled to symmetric forbidden combinations, so one rule covers both directions.
infeasibleList[Dict]default: None
raw forbidden combinations ``[{feature: [values],
monotonicDictdefault: None
``{feature: "increasing" | "decreasing"}`` confirmed constraints applied at training (explicit train_model(monotonic_features=...) wins per feature).
notesDictdefault: None
free-text rationale per entry, e.g. ``{"implies[0]": "no add-ons without an internet plan"}``.
datasets_upload_dataset
Upload a dataset from inline JSON records.
namestrRequired
Name for the dataset
recordsList[Dict]Required
Rows as a list of dicts (JSON records)
descriptionstrdefault: None
Optional description
team_idstrdefault: None
Optional team ID (uses session team_id if not provided)
Preprocessing
preprocessing_add_version_from_spec
Add a new version to an existing preprocessor.
preprocessor_idstrRequired
ID of the existing preprocessor
specDictRequired
PipelineSpec dict
sample_dataList[Dict]default: None
Optional sample data as a list of row dicts (JSON records)
parent_version_idstrdefault: None
Optional parent version for lineage tracking
preprocessing_check_signature
Check whether a preprocessor version's output columns match a model version's training columns (ignoring id and target columns).
preprocessor_version_idstrRequired
The preprocessor version whose output schema to check.
model_version_idstrRequired
The model version whose training columns are the expected schema.
preprocessing_create_preprocessor_from_spec
Create a new preprocessor from a PipelineSpec dict.
namestrRequired
Name of the preprocessor
descriptionstrRequired
Description of the preprocessor
specDictRequired
PipelineSpec dict
sample_dataList[Dict]default: None
Optional sample data as a list of row dicts (JSON records)
preprocessing_get_preprocessor
Get detailed information about a preprocessor.
preprocessor_idstrRequired
ID of the preprocessor
List all available preprocessing transformers with their parameters.
No parameters.
preprocessing_list_preprocessors
List all preprocessors for a team.
team_idstrdefault: None
Optional team ID (uses session team_id if not provided)
preprocessing_preview_from_data
Preview pipeline transformation on sample data.
version_idstrRequired
The version ID to preview
sample_dataList[Dict]Required
Sample data as a list of row dicts (JSON records)
preprocessing_preview_spec
Dry-run a preprocessing spec against a platform dataset without persisting anything.
dataset_idstrRequired
Platform dataset to preview against (see datasets_list_team_datasets)
specDictRequired
PipelineSpec dict: {"version": "2.0", "steps": [{"id", "type", "columns", "params", "description"}]}
target_columnstrdefault: None
Optional prediction target; enables the drop/mutate-label checks
max_rowsintdefault: 10000
Rows to sample for the preview (default 10000)
Deployments
deployments_activate_deployment
Activate a deployment.
deployment_idstrRequired
ID of the deployment to activate
deployments_deactivate_deployment
Deactivate a deployment.
deployment_idstrRequired
ID of the deployment to deactivate
deployments_deploy
Deploy a model version.
model_version_idstrRequired
ID of the model version to deploy
deployments_generate_deploy_key
Generate a deploy key for a deployment.
deployment_idstrRequired
ID of the deployment
descriptionstrdefault: ''
Description of the deploy key use case
days_until_expiryintdefault: 90
Number of days until the key expires
deployments_get_deployment_payload
Get sample payload data for a deployment.
deployment_idstrRequired
ID of the deployment
deployments_list_deployments
List all deployments for a team.
team_idstrdefault: None
Optional team ID (uses session team_id if not provided)
Inference
inference_predict
Predicts the target column for a batch of records.
recordsList[Dict]Required
Rows to score, as JSON records with the model's feature columns (post-preprocessing signature).
model_idstrRequired
The model id (kept for signature compatibility; the route is addressed by version_id).
version_idstrRequired
The version id.
thresholdfloatdefault: 0.5
The threshold for classification models.
inference_score_dataset
Score every row of a platform dataset and return the ranked top-N plus a distribution summary — "who are my highest-risk customers" on a real book.
dataset_idstrRequired
Platform dataset to score (raw / pre-preprocessing columns).
version_idstrRequired
Model version; it must have a deployment (deployments_deploy + deployments_activate_deployment).
thresholdfloatdefault: 0.5
Classification cut-off for `prediction` and the positive counts. Ignored for regression.
top_nintdefault: 50
Rows to return (max 500).
ascendingbooldefault: False
False (default) returns the highest scores first; True the lowest (e.g. safest customers, cheapest quotes).
Optimisers
optimisers_create_optimiser
Create a named optimiser policy on a v2 model version.
model_version_idstrRequired
descriptionstrdefault: None
optimisers_create_optimiser_version
Create a named policy version (OptimizationConfig overrides).
descriptionstrdefault: None
optimisers_list_optimisers
List optimisers for a model.
optimisers_run_optimiser
Run a batch prescriptive optimisation over a dataset.
optimiser_idstrRequired
The optimiser to run.
dataset_idstrRequired
Dataset of rows to optimise (must be deployed model's signature).
version_idstrdefault: None
Optional named policy version; omit for the default policy.
paramsDictdefault: None
Per-run batch kwargs that override the policy. Keys:
run_namestrdefault: None
Optional run label.
run_descriptionstrdefault: None
Optional run description.
optimisers_run_portfolio
Allocate ONE shared intervention budget across every row of a dataset.
model_idstrRequired
The model the optimiser belongs to.
optimiser_idstrRequired
The optimiser (determines the deployed model version).
dataset_idstrRequired
Rows to allocate over.
total_budgetfloatRequired
One shared intervention-cost budget for the whole set.
value_columnstrdefault: None
Optional dataset column used as per-row value weights (revenue, CLV); it is removed from the rows before scoring and switches the objective to value-weighted improvement.
policy_version_idstrdefault: None
Optional named policy version whose saved config (mutable_features, cost_structure, ...) is applied.
paramsDictdefault: None
Per-run overrides. Allowed keys: mutable_features, per_row_immutable (list ALIGNED to rows), feature_bounds, cost_structure, max_joint_candidates, n_grid, cost_resolution.
top_nintdefault: 50
Funded allocations to return (max 500), ranked by improvement.
Reports
reports_create_report
Start async report generation via the wizard.
report_descriptionstrdefault: ''
is_publicbooldefault: False
widgetsList[str]default: None
modestrdefault: 'dynamic'
max_featuresintdefault: 40
constraintsDictdefault: None
audienceDictdefault: None
reports_delete_report
Delete a report and all of its versions.
report_idstrRequired
The ID of the report to delete
reports_get_job_status
Get the status of a wizard report generation job.
Runs
runs_list_team_runs
List the team's runs, newest first.
model_idstrdefault: None
Only runs that produced this model (optional).
team_idstrdefault: None
Team to list for; defaults to the active team.
limitintdefault: 50
Maximum runs to return (newest first).
offsetintdefault: 0
Skip this many, for paging.
AI Reports
gpt_explain_model
Get a natural language explanation of the model.
model_idstrRequired
ID of the model
version_idstrRequired
ID of the model version
languagestrdefault: 'en'
Language for the explanation (e.g., "en", "es", "fr")
detail_levelstrdefault: 'medium'
Level of detail ("low", "medium", "high")
Documentation
docs_get_page
Get the full content of a specific documentation page.
page_idstrRequired
The page identifier, e.g. "getting-started/installation". Use docs_list_pages to discover available page IDs.
docs_list_pages
List all available xplainable documentation pages.
No parameters.
docs_search
Search the xplainable documentation by keyword.
querystrRequired
The search term to look for.
limitintdefault: 5
Maximum number of results to return (default 5).
Utilities
misc_ping_gateway
Ping the API gateway to check connectivity.
hostnamestrdefault: None
Optional hostname to ping (uses session hostname if not provided)
