Atlassian MCP server and Dome
Let agents file the work, and decide what they can rewrite. Jira and Confluence sit behind one Atlassian Rovo MCP server. Connect it to Dome once and each call runs as the person the agent acts for, with rules that read its tool and its site.
Dome provides a tool gatewayand authorization for Atlassian.
One site, by its cloudId
Every Atlassian tool call names a site, and rules refuse any site but yours.
Their login, their permissions
Each engineer authorizes Atlassian from the first call. The agent inherits their Jira and Confluence permissions, and nothing more.
Jira writes, Confluence reads
Rules name the tool. An agent can file and edit work items while every Confluence page stays read-only.
Atlassian behind the Gatewayin three steps.
Add the tools=all endpoint, sync the catalog as yourself, then apply the site rules. Dome registers an OAuth client with Atlassian for you.
Add the Atlassian MCP server
The tools=all endpoint lists every tool by name, so rules can name them. Credentials are per user, through OAuth.
shelldome tools add --name atlassian \ --url "https://mcp.atlassian.com/v2/mcp?tools=all" \ --auth-method oauth --credential-type per-user \ --oauth-authorize-url https://auth.atlassian.com/authorize \ --oauth-token-url https://auth.atlassian.com/oauth/token \ --oauth-registration-url https://auth.atlassian.com/VCeDsk8ZHncYF1g234fKtc4lNipbBhu3/dcr/register \ --oauth-default-scope offline_access \ --oauth-default-scope read:jira:agent-interface \ --oauth-default-scope write:jira:agent-interface \ --oauth-default-scope search:jira:agent-interface \ --oauth-default-scope read:confluence:agent-interface \ --oauth-default-scope search:confluence:agent-interface \ --gateway eng-gatewaySync the catalog
The sync lists tools with your own Atlassian sign-in. Without it, agents see “tool not available in this gateway”.
shelldome tools catalog sync atlassianApply the rules
Start with one agent. Simulate a call against another cloudId and check it's refused before you deploy.
shelldome rules apply atlassian-args.cedar \ --agent incident-scribe --name atlassian-per-site
Your site,and Jira only for writes.
The permit admits the agent. The forbids turn away any cloudId but your site's, and any Confluence create or update.
cedar
permit (principal, action == Dome::Action::"mcp:call", resource is Dome::MCPTool)when { resource.connection_name == "atlassian" };
forbid (principal, action == Dome::Action::"mcp:call", resource is Dome::MCPTool)when { resource.connection_name == "atlassian" && resource has arguments && resource.arguments has cloudId && resource.arguments.cloudId != "7f6c3e2a-1d4b-4c8e-9a0f-2b5d8e1c4a9f"};
forbid (principal, action == Dome::Action::"mcp:call", resource is Dome::MCPTool)when { resource.connection_name == "atlassian" && ["createConfluenceContent", "updateConfluenceContent"].contains(resource.tool_name)};One call,two outcomes.
Switch the caller or the argument and watch the same call decide differently, with every decision landing in audit.
Site
agent incident-scribe · acting as m.reyes
atlassian/createJiraIssue(cloudId: "7f6c3e2a-1d4b-4c8e-9a0f-2b5d8e1c4a9f")
- Agentincident-scribe is registered and active
- Callerm.reyes verified through Okta
- RuleCalls to the acme site are allowed
DecisionAllowed
Three agentson Jira and Confluence.
These come from our template library, and each one reads broadly and writes narrowly.
jira-triager
Triage new work items
Reads each new work item and the ones like it, then sets its labels and assignee.
spec-checker
Check tickets against specs
Reads the Confluence spec behind a work item and comments where the two disagree.
incident-scribe
File incident follow-ups
Reads the runbook and the incident's comments and files the follow-up work in Jira, while the runbook stays with an engineer.
Bringing it alltogether.
Connecting Atlassian to registered agents, models, and identity in Dome completes a governed agent application.
Acting for
Agent
Control point
GatewayEvery call decided and audited
RulesGuardsQuotas
Tools
AtlassianMCP server
Models
Common questionsabout Atlassian.
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
Microsoft Entra ID
Entra access tokens verified on every agent call, so rules read app roles and audit names the person.
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.
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.