Copilot Pre-built Skills¶
Insights Hub Copilot comes with a set of pre-built skills for data access and other basic capabilities. You can use these pre-built skills to create or enhance your own agents. If you require specific capabilities that are not provided by these pre-built skills, you can easily create your own.
This section provides information on the pre-built skills provided and their parameters. The minimum required roles need to be configured when using a technical user and the API.
AnalyzeTimeSeries¶
Description:
This skill analyzes raw or aggregate time series data for one or more variables belonging to the aspect of an asset.
- Performs statistical features (min, max, median, mean, standard deviation, kurtosis, skewness, mode, variance, q25 and q75)
- Calculates the correlation matrix, detects outliers, analyzes trends and detects patterns (dominant period and period score).
- The time range
from,todefaults to 24 hour time range if not provided.
Parameters:
assetIdAspectNameVariables(array): Array of strings in theassetId/aspectName/variablesformat.from(string): Time range start in ISO format.to(string): Time range end in ISO format.timeseriesMode(string): "Raw" or "Aggregate".shouldIncludeDetails(boolean): Include details in the analysis.
Required Roles:
mdsp:core:iot.timUser(Raw time series data)mdsp:core:iot.tsaUser(Aggregated time series data)
CreateCase¶
Description:
This skill creates a new case with the provided details such as title, description, priority, status, type and due date.
Parameters:
description(string): Description of the case.dueDate(string): Due date.priority(string): Priority (EMERGENCY, MEDIUM, HIGH or LOW).status(string): Status (OPEN, INPROGRESS, ON HOLD, DONE, OVERDUE, CANCELLED, ARCHIVED).title(string): Title of the case.type(string): Type (PLANNED, INCIDENT or ANNOTATION).
Required Roles:
mdsp:core:oi.wom.techuser
DeleteCase¶
Description:
This skill permanently deletes an existing case from the system using the case handle as the unique identifier with eTag. If only the case title is known, the GetCases skill can be used to get the required handle. The skill returns a confirmation message or error details.
Parameters:
etag(string): The eTag of the case.handle(string): The unique identifier/handle of the case to be removed from the system, e.g., AA-052, AB-198
Required Roles:
mdsp:core:oi.wom.techuser
FindInDocuments¶
Description:
This skill searches for a given question in documents and gets the answer (RAG approach). The search is performed in the uploaded and synched IDL documents.
Parameters:
question(string): Question to search/find in documents.
Required Roles:
mdsp:core:pcass.r
GetAssetInfo¶
Description:
This skill provides information about the Insights Hub asset model.
- When the optional parameter
nameis provided and one or more matching assets are found, it returns a list of assets. - If the optional parameter
assetIdis provided, it gets details of the asset such as hierarchy path, asset id, name and child assets. The number of child assets can be controlled by using the pagination parameters (pageandsize).
Parameters:
name(string): The name of the asset as a string. Name should be provided with full name, including the path to the asset.assetId(string): Theidof the asset as a string.page(number): The page belongs to pagination and is given as an integer number. Page number starts with 0.size(number): The size belongs to pagination and is given as an integer number. It configures how many items (records) are wanted per page.
Required Roles:
mdsp:core:assetmanagement.reporter
GetCases¶
Description:
This skill gets a filtered list of cases in tabular format. Results are shown in tabular format with only title, description and created date.
Parameters:
title(string): Title of the case.description(string): Description of the case.asset(string): Asset name or ID.priority(string): Priority (EMERGENCY, MEDIUM, HIGH or LOW).status(string): Status (OPEN, INPROGRESS, ON HOLD, DONE, OVERDUE, CANCELLED, ARCHIVED).type(string): Type (PLANNED, INCIDENT or ANNOTATION).duration(string): Duration filter.createdDate(string): Creation date.dueDate(string): Due date.relatedTo(string): Related entities.count(string): Number of cases.sort(string): Sorting mechanism.
Required Roles:
mdsp:core:oi.wom.techuser
GetDocumentsInfo¶
Description:
This skill gets document metadata like names and document URLs regarding the uploaded and synched IDL documents.
Parameters:
None
Required Roles:
mdsp:core:pcass.r
GetEvents¶
Description:
This skill retrieves events filtered by optional parameters.
Parameters:
id(string): A unique identifier of an event.entityId(string): Identifier of the asset.from(string): Start date in ISO 8601 format.to(string): End date in ISO 8601 format.severity(number): Filters events by severity code.acknowledged(boolean): Filters events based on acknowledgment.summarize(boolean): Summarized view.summarizeLimit(number): Threshold for summarization.size(number): Items per page.page(number): Page number for pagination.
Required Roles:
mdsp:core:em.eventviewer
GetTimeInfo¶
Description:
This skill returns a time range or the current time.
- It parses the
timeDescriptionand returnsstartTimeandendTimein ISO format, or the current time. - The parameter
timeDescriptioncontains time information written in natural language. IftimeDescriptionis not set, 'now' is used as the default.
Parameters:
timeDescription(string): Contains time information written in natural language, in English.
Required Roles:
- no role required
GetTimeSeriesAggregates¶
Description:
This skill gets aggregated time series data for the provided asset id, aspect name, and variables for the given duration or for the to and from date.
- If
fromandtoare not specified, it returns data for the last 24 hours.
Parameters:
assetId(string): Asset id.aspectName(string): Aspect name.from(string): Start time in ISO format.to(string): End time in ISO format.variables(string): Comma-separated list of variables.
Required Roles:
mdsp:core:iot.tsaUser(Aggregated time series data)
GetTimeSeriesRaw¶
Description:
This skill gets raw time series data for the provided asset id, aspect name, and variables for the given duration or for the to and from date.
- If
fromandtoare not specified, it returns data for the last 24 hours.
Parameters:
assetId(string): Asset id.aspectName(string): Aspect name.from(string): Start time in ISO format.to(string): End time in ISO format.variables(string): Comma-separated list of variables.
Required Roles:
mdsp:core:iot.timUser(Raw time series data)
ModelReadOps¶
This skill reads the model registry and serving state. It supports listing and fetching models, versions, artifacts, servers, and server status. All operations are non-destructive.
- Use
list_modelsto browse all registered models, with optional filtering, sorting, and pagination. - Use
get_model_by_idorget_model_by_nameto fetch a specific model. - Use
list_versionsto list all versions of a given model. - Use
get_version_by_idorget_version_by_nameto fetch a specific version. - Use
list_artifactsto list all artifacts belonging to a version. - Use
get_artifact_by_idto fetch a specific artifact. - Use
list_serversto browse all deployed model servers. - Use
get_server_statusto check the current status of a deployed server.
Parameters:
| Parameter | Type | Description |
|---|---|---|
action | string | Operation to perform. Allowed values: list_models, get_model_by_id, get_model_by_name, list_versions, get_version_by_id, get_version_by_name, list_artifacts, get_artifact_by_id, list_servers, get_server_status. |
model_id | string | Unique ID of the model. Required for: list_versions, get_model_by_id. |
name | string | Exact name of the model or version. Required for: get_model_by_name, get_version_by_name. |
version_id | string | Unique ID of the version. Required for: get_version_by_id, list_artifacts. |
artifact_id | string | Unique ID of the artifact. Required for: get_artifact_by_id. |
id | string | Unique ID of the model server. Required for: get_server_status. |
filterQuery | string | Optional filter expression for list actions. |
pageSize | number | Optional number of results per page. Integer range: 1 to 1000. |
nextPageToken | string | Optional continuation token for paged list actions. |
sortBy | string | Optional field used to sort results. Applicable to: list_models. |
sortOrder | string | Optional sort direction. Allowed values: ASC, DESC. Applicable to: list_models. |
requestId | string | Optional caller-supplied correlation ID. |
ModelWriteOps¶
This skill creates and updates models, versions, and artifacts in the model registry. All create operations follow a human-in-the-loop confirmation flow — the agent collects inputs, presents a payload preview, and waits for explicit user confirmation before submitting.
- Use
create_modelto register a new model. - Use
update_modelto update the name, description, or metadata of an existing model. - Use
create_versionto register a new version under an existing model. - Use
update_versionto update the description or metadata of an existing version. - Use
create_artifactto register a new artifact (model file, metric, parameter, dataset, or document) under an existing version. - Use
update_artifactto update fields of an existing artifact. - Use
create_model_version_artifactto register a model, a version, and an artifact in a single call.
Note
Before registering an artifact, the file must be physically present in the data lake at the exact storagePath provided. This skill registers a reference to the path — it does not upload or validate the file's existence.
Parameters:
| Parameter | Type | Description |
|---|---|---|
action | string | Operation to perform. Allowed values: create_model, update_model, create_version, update_version, create_artifact, update_artifact, create_model_version_artifact. |
name | string | Name of the resource. Required for: create_model, create_version, create_artifact, create_model_version_artifact. |
origin | string | Owning system or source, for example IDL. Required for: create_model, create_model_version_artifact. |
model_id | string | Unique ID of the model. Required for: update_model, create_version. |
version_id | string | Unique ID of the version. Required for: update_version, create_artifact. |
artifact_id | string | Unique ID of the artifact. Required for: update_artifact. |
artifactType | string | Type of artifact. Required for: create_artifact, create_model_version_artifact. Allowed values: model-artifact, metric, parameter, dataset-artifact, doc-artifact. A version must include at least one model-artifact to be deployable. |
storagePath | string | Data lake path to the artifact file, for example data/ten=demo/models/MyModel/model.bst. Required for: create_artifact, update_artifact, create_model_version_artifact. |
modelFormatName | string | Model format name. Required for: create_artifact, update_artifact, create_model_version_artifact. Allowed values: sklearn, tensorflow, pytorch, onnx, xgboost, lightgbm, keras, h5, pmml, custom. |
resources | object | CPU and memory allocation for the artifact, for example {"cpu": "0.5", "memory": "1Gi"}. Required for: create_artifact, create_model_version_artifact. |
description | string | Optional human-readable description. |
uri | string | Optional URI reference. |
modelFormatVersion | string | Optional version of the model format. |
storageKey | string | Optional storage key. |
customProperties | object | Optional key-value metadata in ML Metadata wrapper format, such as MetadataStringValue, MetadataIntValue, MetadataDoubleValue, and MetadataBoolValue. |
model_data | object | Required for create_model_version_artifact. Contains model fields: name, origin (required), and description (optional). |
model_version_data | object | Required for create_model_version_artifact. Contains version fields: name (required) and description (optional). |
model_artifact_data | object | Required for create_model_version_artifact. Contains artifact fields: name, artifactType, storagePath, modelFormatName, and resources (required), plus description, uri, storageKey, and modelFormatVersion (optional). |
requestId | string | Optional caller-supplied correlation ID. |
ModelExecutionOps¶
This skill executes runtime operations against the model registry and serving infrastructure. It supports deploying a model version to a serving runtime, comparing two versions of the same model side-by-side, and running inference against a deployed model server.
- Use
deployto deploy a specific model version. A pre-flight check confirms that amodel-artifactexists on the version before deployment is triggered. - Use
compare_versionsto compare two versions of the same model and view their differences. - Use
run_inferenceto send an inference request to a deployed model server and retrieve the prediction result.
Parameters:
| Parameter | Type | Description |
|---|---|---|
action | string | Operation to perform. Allowed values: deploy, compare_versions, run_inference. |
model_id | string | Unique ID of the model. Required for: deploy, compare_versions. |
version_id | string | Unique ID of the version to deploy. Required for: deploy. |
version_id_a | string | Unique ID of the first version to compare. Required for: compare_versions. |
version_id_b | string | Unique ID of the second version to compare. Required for: compare_versions. |
server_id | string | Unique ID of the deployed model server. Required for: run_inference. |
input_data | object | Inference payload as a valid JSON object that matches the model's expected input schema. Required for: run_inference. |
requestId | string | Optional caller-supplied correlation ID. |
* assetId (string): Asset id. | ||
* aspectName (string): Aspect name. | ||
* from (string): Start time in ISO format. | ||
* to (string): End time in ISO format. | ||
* variables (string): Comma-separated list of variables. |
Required Roles:
mdsp:core:iot.timUser(Raw time series data)