Announcing Dome Managed Agents.

Try it now

Ollama and Dome

Models on your own hardware with Ollama, governed like any provider. Ollama runs models on a workstation or a server you own. Dome treats it as one more provider, so its calls get the same rules and audit as a hosted model. The one requirement: the Gateway must be able to reach the Ollama URL.

Dome provides model brokeringand routing for Ollama.

Any model Ollama runs

Use Dome's openai_compatible provider with Ollama's /v1 address. The model tag you pulled is the model id.

A second server behind it

Pool Ollama with a vLLM server serving the same model. When Ollama fails, the call goes to vLLM.

Development stays in development

Tag Ollama connections env: dev. A rule on each production agent refuses them.

Connect Ollamain three steps.

Connect the server, expose it to the Gateway, and pool a second server if you run one.
  1. Add the connection

    Use the provider id openai_compatible and Ollama's /v1 URL. Ollama ignores API keys, so set no credential.

    shell
    dome models add gpt-oss-ollama \  --provider openai_compatible \  --endpoint http://ollama.internal.example.com:11434/v1 \  --model gpt-oss:20b \  --auth-method none --credential-type none \  --attributes '{"env":"dev"}' \  --gateway dev-gateway
  2. Expose it to the Gateway

    Dome's hosted gateway can't see localhost or a private network. Ollama needs an address it can reach.

  3. Pool it and call the Gateway

    Ollama goes first and vLLM second. Your client calls gpt-oss-20b with the agent's Dome key.

    shell
    dome models pool create gpt-oss-20b \  --failover-max all --gateway dev-gateway
    dome models pool member add gpt-oss-20b gpt-oss-ollama --priority 0dome models pool member add gpt-oss-20b gpt-oss-vllm --priority 1

Production agentsstay off dev models.

Applied to a production agent with --agent, this refuses any connection tagged env: dev, so a workstation model never answers a customer.
cedar
forbid (principal, action == Dome::Action::"llm:invoke", resource is Dome::LLMModel)when { resource has env && resource.env == "dev" };

Bringing it alltogether.

Connecting Ollama to registered agents, tools, and identity in Dome completes a governed agent application.
Dome

Control point

GatewayEvery call decided and audited
RulesGuardsQuotas

Common questionsabout Ollama.

How does Dome connect to Ollama?
Dome's openai_compatible provider, pointed at Ollama's /v1 API. No API key, since Ollama doesn't check one.
Can a Dome-hosted gateway reach Ollama on my laptop?
No. It can't reach localhost or a private network, so expose Ollama on a URL the gateway can reach.
Which model names do I use?
Ollama's own tags, such as gpt-oss:20b or qwen3:8b. Dome passes the tag through unchanged.

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

LangGraph

LangGraph agents with their model calls on the Model Broker and their MCP tools on the Tool Gateway.
Client

Codex

Codex 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.

Every agent you run, under one system of control.