Announcing Dome Managed Agents.

Try it now

Linear MCP server and Dome

Let agents file issues, and decide which teams they land in. Linear is where agents file the work they find. Connect its MCP server to Dome once, and every issue an agent saves is checked against the team it names, with the Linear login of the person it acts for.

Dome provides a tool gatewayand authorization for Linear.

Decided on the team

One tool, allowed in one team and refused in another. Rules read the team of every issue an agent saves.

Their login, their workspace

An engineer authorizes Linear the first time an agent calls it for them. From then on the agent sees the teams that engineer sees.

Comments added, never deleted

Rules name the tool. An agent can comment on any issue while delete_comment is refused.

Linear behind the Gatewayin three steps.

Add Linear's server by URL, sync its tools once you've signed in, then scope the team rules to one agent. Dome brings its own OAuth client.
  1. Add the Linear MCP server

    Credentials are per user, through OAuth. Each engineer consents on their first call.

    shell
    dome tools add --name linear \  --url https://mcp.linear.app/mcp \  --auth-method oauth --credential-type per-user \  --oauth-authorize-url https://mcp.linear.app/authorize \  --oauth-token-url https://mcp.linear.app/token \  --oauth-registration-url https://mcp.linear.app/register \  --oauth-default-scope read --oauth-default-scope write \  --gateway eng-gateway
  2. Sync the catalog

    The sync calls Linear as you, so finish your own sign-in first. Skip it and agents get “tool not available in this gateway”.

    shell
    dome tools catalog sync linear
  3. Apply the rules

    Try the Support rule on one agent with --agent, and simulate a save_issue for another team before you deploy.

    shell
    dome rules apply linear-args.cedar \  --agent support-triager --name linear-per-team

Supportteam only.

The permit gives the agent Linear. One forbid refuses a save_issue for any team but Support, and refuses a team given by ID, and the other refuses delete_comment.
cedar
permit (principal, action == Dome::Action::"mcp:call", resource is Dome::MCPTool)when { resource.connection_name == "linear" };
forbid (principal, action == Dome::Action::"mcp:call", resource is Dome::MCPTool)when {  resource.connection_name == "linear" &&  resource.tool_name == "save_issue" &&  resource has arguments && resource.arguments has team &&  resource.arguments.team != "Support"};
forbid (principal, action == Dome::Action::"mcp:call", resource is Dome::MCPTool)when {  resource.connection_name == "linear" &&  resource.tool_name == "delete_comment"};

One call,two outcomes.

Switch the caller or the argument and watch the same call decide differently, with every decision landing in audit.

Team

agent support-triager · acting as r.haddad
linear/save_issue(team: "Support", title: "SSO login loops for Okta users")
  1. Agentsupport-triager is registered and active
  2. Callerr.haddad verified through Okta
  3. RuleIssues in Support are allowed
DecisionAllowed

Three agentson Linear.

These come from our template library, and each one reads broadly and writes narrowly.

support-triager

File support issues

Turns support conversations into Support team issues with priority and labels set, and is refused if it moves one to another team.

bug-deduper

Flag duplicate bugs

Searches every team for duplicates of each new bug and comments with the likely match, while closing stays with an engineer.

changelog-writer

Draft the changelog

Reads the issues completed this cycle and drafts the changelog as a Linear document.

Bringing it alltogether.

Connecting Linear to registered agents, models, and clients in Dome completes a governed agent application.
Dome

Control point

GatewayEvery call decided and audited
RulesGuardsQuotas

Tools

LinearMCP server
GitHubMCP server

Common questionsabout Linear.

Does Dome work with Linear's remote MCP server?
Yes. Add mcp.linear.app by URL with OAuth. Dome registers its own client, and each engineer authorizes on their first call.
Can I limit which Linear teams an agent files issues in?
Yes. Rules read the team argument of save_issue, so the same tool is allowed in one team and refused in another.
Can an agent comment on issues without deleting comments?
Yes. Rules name the tool, so save_comment is allowed while delete_comment is refused.
Whose Linear permissions does an agent use?
Those of the person it acts for. Each engineer's Linear token is held in Dome and chosen by their verified identity, so an agent never reaches a team they can't.

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.
Provider

Anthropic

The Claude API behind the Model Broker: the key held in Dome, every call authorized, metered and audited.
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.

MCP servers for AI agents: per-argument rules on every call

Put any remote MCP server behind the Tool Gateway. Agents reach it through one governed endpoint, and rules decide each call on its arguments.

Every agent you run, under one system of control.