Databricks Genie Code
BetaDatabricks' autonomous agent for data engineering, analytics and ML, embedded in the workspace, with Agent Skills support.
- AI Agent
- Data
- Cloud
- Authentication
- OAuth 2.0
- Required scopes
- 3
How Databricks Genie Code works with Ferentin
Genie Code adds an external server through a Unity Catalog HTTP connection, which supports Bearer tokens, OAuth machine-to-machine, per-user OAuth and Dynamic Client Registration. On the DCR path Databricks drives the whole handshake itself — discovering the authorization server and registering its own OAuth credentials — so connecting it to a Ferentin MCP endpoint needs no client id copied by hand. Because it runs server-side rather than on a developer's machine, it is the one client here with no endpoint footprint: Ferentin governs its MCP traffic without detecting it on a device.
Its manually-configured OAuth forms require a client secret, which the AI-agent wizard does not yet mint — use the DCR path, or create a confidential client by hand.
- 1
Register Databricks Genie Code
Add Databricks Genie Code as an AI application in the Ferentin admin console.
- 2
Sign in with your identity
People authorize Databricks Genie Code with their enterprise identity, over OAuth.
- 3
Every tool call governed
Ferentin checks each MCP call against your policies and logs it.
What you get
Agent skills
Reusable skills extend what the agent knows how to do.
MCP client
Databricks Genie Code calls MCP servers, so it can reach your tools through Ferentin.
Dynamic client registration
Databricks Genie Code registers itself with Ferentin, so there is no client secret to hand out.
Data analysis
Explore and chart data in a conversation.
Cloud agent
Agents run in the cloud and keep working after you close the laptop.
Permissions requested
The OAuth 2.0 scopes Databricks Genie Code asks you to approve when you connect.
| Scope | Access |
|---|---|
| openid | Not rated |
| profile | Not rated |
| mcp | Not rated |
Servers Databricks Genie Code can reach through Ferentin
Browse all 112 MCP servers
Salesforce
Agent Browser
Sentry
Agent Terminal
Atlassian
AWS
Box
GitHubAbout Databricks Genie Code
Genie Code is an AI agent built into the Databricks platform and purpose-built for data work — data engineering, analytics, machine learning and agent building. Skills are workspace files rather than files on a developer's machine, managed from the Genie Code pane and version-controlled through Databricks Git folders.
Frequently asked about Databricks Genie Code
- How does Databricks Genie Code work with Ferentin?
- Ferentin registers Databricks Genie Code as an AI application. People sign in with their enterprise identity, and every MCP tool call Databricks Genie Code makes is checked against your Ferentin policies and logged.
- What permissions does the Databricks Genie Code integration require?
- The Databricks Genie Code integration requests 3 OAuth 2.0 scopes: openid; profile; mcp
- How do I connect Databricks Genie Code to Ferentin?
- Open the Ferentin admin console and register Databricks Genie Code as an AI application. Then point Databricks Genie Code at your Ferentin MCP endpoint and sign in.
- Is the Databricks Genie Code integration generally available?
- The Databricks Genie Code integration is currently beta on Ferentin.
