Neptune DXP Documentation
Neptune DXP - Open Edition · 25.0 MTS

AI agents, tools, and models

AI tools maintain the agents, tools, and models of Naia Agent Studio. An agent uses tools and a model, so create or update the tools and the model before the agent that refers to them. To build a valid model configuration, call list_ai_vendor_settings first.

Required role: ai-agent, ai-tool, ai-model.

If your user account does not have this role, every tool in this group fails immediately with an access denied error. Contact an administrator to request the role, or see Access control for more information on how roles are enforced.

To keep responses small, related artifacts are reduced to their id and name in the responses of these tools. Use the tools of the related artifact type to read the full object.

Available tools

Parameters are shown with their type and a short description. UUIDs are always strings. The list_* tools accept listOptions for filtering, field selection, and sorting on the fields named for each tool.

AI agents

The agent tools work with standard (local) agents. External agents, such as A2A peers, are registered separately. save_ai_agent returns an error if you try to create an external agent.

In the responses of get_ai_agent and save_ai_agent, the assigned tools, apps, launchpads, peers, and sapDataSources, and the scripts, apps, launchpads, processFlows, and peers listed under whereUsed are reduced to id and name.

Tool Description Parameters

list_ai_agents

List AI agents.

listOptions (object, optional). Fields: name, model, description, enableIntelligentApps, version, package, createdBy, changedBy, createdAt, updatedAt.

get_ai_agent

Return an AI agent by its ID, including its configuration, prompts, roles, and tools, and the artifacts that it is assigned to or used by.

id (string): ID of the AI agent.

save_ai_agent

Create or update a standard AI agent. Pass the full agent object, including the model and the configuration. Omit id to create a new agent.

aiAgent (object): Full AI agent definition object.

delete_ai_agent

Delete an AI agent by its ID.

id (string): ID of the AI agent.

The field enableIntelligentApps corresponds to the Enable use with Agentic Apps setting of an agent.

AI tools

In the responses of list_ai_tools, get_ai_tool, and save_ai_tool, the roles of a tool are reduced to id and name.

Tool Description Parameters

list_ai_tools

List AI tools. Without field selection, each entry contains id, name, type, package, and updatedAt.

listOptions (object, optional). Fields: name, type, description, version, package, createdBy, changedBy, createdAt, updatedAt.

get_ai_tool

Return an AI tool by its ID, including the agents it is assigned to.

id (string): ID of the AI tool.

save_ai_tool

Create or update an AI tool. Pass the full tool object, including the tool type and its type-specific configuration. Omit id to create a new tool.

aiTool (object): Full AI tool definition object.

delete_ai_tool

Delete an AI tool by its ID.

id (string): ID of the AI tool.

AI models

In the responses of the model tools, the roles of a model and the agents, tables, and guardrails listed under whereUsed are reduced to id and name.

Tool Description Parameters

list_ai_models

List AI models. The vendor configuration (config) is never included in the list, even if you select it. Use get_ai_model to read it.

listOptions (object, optional). Fields: name, type, description, version, inputType, outputType, package, createdBy, changedBy, createdAt, updatedAt.

get_ai_model

Return an AI model by its ID, including its vendor configuration and where it is used.

id (string): ID of the AI model.

save_ai_model

Create or update an AI model. Pass the full model object, including the vendor type and the configuration. Omit id to create a new model.

aiModel (object): Full AI model definition object.

delete_ai_model

Delete an AI model by its ID. The deletion fails if an agent uses the model or if the model is set as the global embedding model.

id (string): ID of the AI model.

list_ai_vendor_settings

List every AI model vendor that the system supports. For each vendor, the response contains the names of its configuration fields, for example, apiKey, region, or resourceName, the available embedding models with their vector dimensions, and the available image generation models. Only field names are returned, no configured values.

None

Related topics