skills.rest
FMurray
FMurrayCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

agent-bricks

Create and manage Databricks Knowledge Assistants, Genie Spaces, and Multi-Agent Supervisors.

Updated Mar 5, 2026
Search Tags:#rag#conversational ai#databricks#knowledge assistant#genie space#agent bricks#multi-agent supervisor
One-click install
npx skills add https://github.com/FMurray/mlfts --skill agent-bricks-fmurray
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: agent-bricks
Source: https://github.com/FMurray/mlfts/tree/main/.agents/skills/agent-bricks
Command: npx skills add https://github.com/FMurray/mlfts --skill agent-bricks-fmurray

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Building conversational AI applications on Databricks requires coordinating document indexing, SQL interfaces, and multi-agent routing, which involves many manual configuration steps. This Skill provides structured guidance and MCP tool references for creating and managing Databricks Agent Bricks end to end. ## Core Features & Use Cases - Knowledge Assistants (KA): Create RAG-based document Q&A systems over PDF and text files stored in Unity Catalog Volumes, with automatic example ingestion from companion JSON files. - Genie Spaces: Set up natural-language-to-SQL exploration over Unity Catalog tables, with detailed guidance delegated to the databricks-genie skill. - Multi-Agent Supervisors (MAS): Orchestrate multiple specialized agents (KA endpoints, Genie spaces, custom model serving endpoints) behind a single routing endpoint, including hierarchical MAS architectures. - Use Case: Generate synthetic HR policy PDFs, create a Knowledge Assistant over them, then combine it with a Genie space for usage analytics inside a Multi-Agent Supervisor that routes employee questions to the right agent. ## Quick Start Ask the agent to create a Knowledge Assistant named HR Policy Assistant over the volume /Volumes/my_catalog/my_schema/raw_data/hr_docs and wait for its endpoint to reach ONLINE status.

Frequently Asked Questions about agent-bricks

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I create a Knowledge Assistant on Databricks?▼

Use the create_or_update_ka tool with a name, a volume_path pointing to documents in a Unity Catalog Volume, and optional description and instructions. The endpoint provisions in 2-5 minutes and reaches ONLINE status when ready.

How do I route queries across multiple agents with a Multi-Agent Supervisor?▼

Call create_or_update_mas with a list of agents, each having a name, a specific description, and exactly one of ka_tile_id, genie_space_id, or endpoint_name. Clear, non-overlapping agent descriptions are critical for accurate routing.

Can I use a Genie space as an agent in a Multi-Agent Supervisor?▼

Yes, Genie spaces can be added directly as MAS agents using their genie_space_id, with no separate endpoint deployment needed. Use find_genie_by_name to look up the space ID, since there is no system table for Genie spaces.

How do I find a Knowledge Assistant endpoint name by its name?▼

Use the find_ka_by_name tool with the exact KA name. It returns the tile_id, sanitized name, endpoint_name following the pattern ka-{tile_id}-endpoint, and the current endpoint_status.

Why is my Knowledge Assistant stuck in PROVISIONING status?▼

Provisioning typically takes 2-5 minutes, so wait up to 10 minutes before investigating. If it persists, check workspace capacity and quotas, and verify the volume path is accessible and correctly specified.

What data do I need before creating Agent Bricks?▼

Knowledge Assistants need PDF or text documents in a Unity Catalog Volume, which can be generated with the unstructured-pdf-generation skill. Genie Spaces need tables in Unity Catalog, which can be built with synthetic-data-generation and spark-declarative-pipelines skills.

Related Skills

Explore More Data & Analytics Skills→
leary-pokenleary-poken
1

databricks-unstructured-pdf-generation

Generates realistic PDFs and JSON metadata for ML-formatting purposes.

Community
Intermediate
datasciencemonkeydatasciencemonkey
27

databricks-agent-bricks

Automates provisioning of document, SQL and hybrid AI assistants via customizable templates.

Community
Advanced
BlackkadderBlackkadder

databricks-model-serving

Deploy MLflow models and GenAI agents to scalable REST API endpoints.

Community
Advanced
pascalvanderheidenpascalvanderheiden

rag-patterns

Build RAG pipelines with Azure AI Search for hybrid retrieval.

Community
Advanced