Databricks Model Serving and Dome
Your Databricks endpoints, with every agent call governed. Databricks Model Serving gives each workspace its own endpoints. Connect those endpoints to Dome and a rule keeps production agents on production workspaces. Anthropic's API stands behind Claude Opus 5.5 if Databricks fails.
Dome provides model brokeringand routing for Databricks.
The token never leaves Dome
Dome stores the Databricks token and presents it per call. Agents never get a workspace credential.
Databricks first, the vendor behind it
Pool the Databricks endpoint for Claude Opus 5.5 with Anthropic's API. When Databricks fails, the call goes to Anthropic.
Production agents on production workspaces
Tag each connection with its workspace's environment. A rule keeps a production agent off dev endpoints.
Connect Databricksin three steps.
Add each serving endpoint, pool it with the vendor's API, and change one base URL in each agent.
Add the connection
The provider id is databricks. Databricks endpoints are per workspace, so set --endpoint to your workspace URL plus /serving-endpoints; the model is the serving endpoint's name.
shelldome models add opus-databricks \ --provider databricks \ --endpoint https://<workspace>.cloud.databricks.com/serving-endpoints \ --model databricks-claude-opus-5-5 \ --api-key "$DATABRICKS_TOKEN" \ --attributes '{"env":"prod"}' \ --gateway prod-gatewayPool it with Anthropic
Add an Anthropic connection for claude-opus-5-5, tagged env: prod too, and pool the two so agents send the pool's name.
shelldome models pool create claude-opus-5-5 \ --failover-max all --gateway prod-gateway dome models pool member add claude-opus-5-5 opus-databricks --priority 0dome models pool member add claude-opus-5-5 opus-anthropic --priority 1Point your agent at the Gateway
Point the OpenAI SDK at the Gateway with the agent's Dome key, and call the pool by name.
pythonfrom openai import OpenAI client = OpenAI(base_url=f"{GATEWAY_URL}/v1", api_key=DOME_AGENT_KEY)client.chat.completions.create( model="claude-opus-5-5", messages=[{"role": "user", "content": "Explain why this pipeline run failed."}],)
Production agentsuse production endpoints.
Applied to a production agent with --agent, this refuses any connection not tagged env: prod. A dev workspace's endpoint is out of reach.
cedar
forbid (principal, action == Dome::Action::"llm:invoke", resource is Dome::LLMModel)unless { resource has env && resource.env == "prod" };Bringing it alltogether.
Connecting Databricks to registered agents, tools, and identity in Dome completes a governed agent application.
Acting for
Agent
Control point
GatewayEvery call decided and audited
RulesGuardsQuotas
Models
Common questionsabout Databricks.
More of whatDome works with.
Model
Claude Fable
Fable 5.1 from Anthropic and Amazon Bedrock in one failover pool, open to one group and capped by quota.
MCP server
GitHub
The GitHub MCP server behind the Tool Gateway, with rules that decide each call on its owner and repository.
Identity
Okta
Okta tokens verified on every agent call, so rules and audit name the person each agent acted for.
Runtime
OpenAI Agents SDK
OpenAI Agents SDK agents with their models on the Model Broker and their MCP tools on the Tool Gateway.
Client
Claude Code
Claude Code on a Dome Gateway with per-developer sign-in, rules on every tool call, and audit by name.
Agent service
TinyFish
TinyFish's web agents behind the Tool Gateway, with rules that decide each run on the site it targets.
Talk with ourFDE team.
Or start free, and register your first agent in minutes. Connect it through a gateway, give it rules, and see every call it makes in one audit trail.
Talk with our FDE team
Our engineers design, build and run governed agents for the part of your business you want to change. You see what they deliver.
LLM providers for AI agents: one governed path to every model
Connect a provider once. Its key stays in Dome, and every agent call to it is authorized, metered and audited.