Skip to main content
Version: v4.0.0 [Denim]

MX-AI Catalog

Agents Catalog

SMO Agent - Network Management and Operations - (DEPRECATED)

This agent is deprecated and will be removed in future releases. Please use the SMO Agent V2 whenever possible.

Image: hub.bubbleran.com/orama/agents/smo-agent

Functionalities:

  • Leverages the Observability DB MCP Server to provide observability capabilities:
    • Monitor the deployed Networks and UEs and their configurations.
    • Count/List resources.
  • Deploys new Networks and UEs allowing to specify a minimal configuration:
    • Network: Network name, Access Networks number and names, whether to include FlexRIC or not in the deployment.
    • Terminal (UE): UE name, preferred Access Network, IMSI (optional).

SMO Agent V2 - Network Management and Operations

Image: hub.bubbleran.com/orama/agents/smo-agent-v2

Functionalities:

  • Leverages the Observability DB MCP Server to provide observability capabilities:
    • Monitor the deployed Networks and UEs and their configurations.
    • Count/List resources.
  • Leverages Hermes to deploy new Networks and UEs.
  • Leverages Hermes to update existing Networks and UEs.
  • Leverages Hermes to delete existing Networks and UEs.

Hermes - Kubernetes API Agent

Image: hub.bubbleran.com/orama/agents/hermes

Functionalities:

  • Translates natural language intents into Kubernetes API calls.
  • Validates the API request based on the available schema.
  • Tested and validated with Athena APIs. Other APIs will be tested for future releases.

Supervisor - Agents Orchestrator - (DEPRECATED)

This agent is deprecated and will be removed in future releases. Please use the Supervisor V2 whenever possible.

Image: hub.bubbleran.com/orama/agents/supervisor

Functionalities:

  • Leverages other agents to translate human intents into a well-structured plan.
    • Each step of the plan is executed by another agent.
    • The Supervisor puts together all of the results to generate a final answer for the user.
  • Handles Human-In-The-Loop (HITL) interactions.

Supervisor V2 - Agents Orchestrator

Image: hub.bubbleran.com/orama/agents/supervisor-v2

Functionalities: same as Supervisor, but this version is more stable and compliant with the latest version of BAT-ADK.

UL Anomaly Detection Agent - Detect UL interference from RAN KPIs - (EXPERIMENTAL)

Image: hub.bubbleran.com/orama/agents/ul-ad-agent

Functionalities:

  • Leverages the Anomalies DB MCP Server to retrieve anomaly alerts generated by the inference pipeline.
  • Assesses network status and anomaly severity based on anomaly scores and deviating KPIs.
  • Uses KPI context and explanations to provide anomaly interpretation, root cause analysis, and recommended actions.

P0 Configuration Agent - Control P0 Nominal to optimize UL throughput - (EXPERIMENTAL)

Image: hub.bubbleran.com/orama/agents/p0-agent

Functionalities:

  • Provides observability on UE power-related analytics, including p0Nominal PUSCH, PUCCH, SNR and DL/UL throughput.
  • Employes 2 tools, KPI_mon() and p0_enforce(), in order to collect the metrics and enforce new p0Nominal configurations.
  • Can use tools, reason and propose improvements on the p0Nominal configuration.

MCP Servers Catalog

Observability DB

Image: hub.bubbleran.com/orama/mcp/observability-db

Functionalities:

  • Collect Kubernetes resources, split them into chunks and save into a Vector Database.

Tools:

  • Query: perform a similarity search between an input query and the retrieved resources.
  • Enumerate: perform an enumerate operation on the retrieved resources.
    • Currently supported operations: count, list.

Anomalies DB - (EXPERIMENTAL)

Image: hub.bubbleran.com/orama/mcp/anomaly-db

Functionalities:

  • Collects and stores real-time anomaly alerts produced by the inference pipeline.
  • Maintains anomaly metadata, including anomaly scores and deviating KPIs.

Tools:

  • get_alerts_in_interval: Query anomalies over a user-defined time interval.
  • get_kpi_guides: Provides KPI contextual information to support explainability and reasoning by agents.