Sentry MCP server and Dome
Let agents triage Sentry issues without letting them silence one for good. Connect Sentry's MCP server to Dome with OAuth. Each engineer consents once, and calls made for them run with their Sentry access. Rules read the arguments of every call, down to how an issue is ignored.
Dome provides a tool gatewayand authorization for Sentry.
Decided on the arguments
The same update_issue call resolves an issue or ignores it until it escalates, but ignoring it forever is refused.
Their Sentry access
With per-user OAuth, each engineer consents once. An agent acting for them reaches only what they can.
No side door
Sentry's execute_sentry_tool runs any tool in its catalog by name. Rules read that name too, so deletes routed through it are refused.
Sentry behind the Gatewayin three steps.
Add the server, sync its tools into the Gateway's catalog, and apply the rules. Sentry supports dynamic client registration, so there is no OAuth app to create.
Add the Sentry MCP server
Dome registers itself as a client with Sentry and holds each engineer's tokens. Calls made for them carry their own.
shelldome tools add --name sentry \ --url https://mcp.sentry.dev/mcp \ --auth-method oauth --credential-type per-user \ --oauth-authorize-url https://mcp.sentry.dev/oauth/authorize \ --oauth-token-url https://mcp.sentry.dev/oauth/token \ --oauth-registration-url https://mcp.sentry.dev/oauth/register \ --gateway eng-gatewaySync the catalog
Run it once your own Sentry consent is attached. Until then the sync fails, and agents get “tool not available in this gateway”.
shelldome tools catalog sync sentryApply the rules
Scope them to one agent while you try them, and simulate before you deploy.
shelldome rules apply sentry-triage.cedar \ --agent issue-triager --name sentry-triage
Triage freely,never ignore forever.
The permit opens Sentry to the agent. The first forbid refuses ignoring an issue forever. The second refuses deletes, and update_issue itself, when they come through execute_sentry_tool.
cedar
permit (principal, action == Dome::Action::"mcp:call", resource is Dome::MCPTool)when { resource.connection_name == "sentry" };
forbid (principal, action == Dome::Action::"mcp:call", resource is Dome::MCPTool)when { resource.connection_name == "sentry" && resource.tool_name == "update_issue" && resource has arguments && resource.arguments has ignoreMode && resource.arguments.ignoreMode == "forever"};
forbid (principal, action == Dome::Action::"mcp:call", resource is Dome::MCPTool)when { resource.connection_name == "sentry" && resource.tool_name == "execute_sentry_tool" && resource has arguments && resource.arguments has name && (resource.arguments.name like "delete_*" || resource.arguments.name == "update_issue")};One call,two outcomes.
Switch the caller or the argument and watch the same call decide differently, with every decision landing in audit.
Ignore mode
agent issue-triager · acting as j.alvarez
sentry/update_issue(organizationSlug: "acme", issueId: "PAYMENTS-4F2", status: "ignored", ignoreMode: "untilEscalating")
- Agentissue-triager is registered and active
- Callerj.alvarez verified through Okta
- RuleIssues may be ignored until they escalate
DecisionAllowed
Three agentson Sentry.
These come from our template library, and each one reads broadly and writes narrowly.
issue-triager
Triage new issues
Reads new issues and their events, then resolves, assigns, or ignores them until they escalate.
root-cause-analyst
Find the root cause
Pulls the stack trace and surrounding events for an issue and asks Seer for the root cause without changing anything in Sentry.
alert-tuner
Tune noisy alerts
Finds alert rules that fire too often and adjusts their thresholds, while deleting a rule stays with an engineer.
Bringing it alltogether.
Connecting Sentry 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
SentryMCP server
Models
Common questionsabout Sentry.
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
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.