
Grafbase
Your AI agent monitors schemas, reviews federation checks, manages API keys, and inspects subgraphs in your Grafbase project. Developers get instant answers about their GraphQL infrastructure without leaving the chat. DevOps visibility at conversation speed.




Your AI agent navigates Grafbase's schema registry, federation layer, and API key management, giving developers instant visibility into their GraphQL platform.
Grafbase
Real scenarios where developers and platform teams use AI agents to manage schemas, audit API access, and monitor their Grafbase federation without touching dashboards.
A developer is about to deploy a schema change and wants to confirm the latest check passed. The AI Agent retrieves the most recent schema check from Grafbase, confirms it passed with no breaking changes, and lists the affected subgraphs. The developer gets deployment confidence from a single chat message instead of navigating the dashboard and CLI output.
A security audit reveals an API key needs to be rotated. The AI Agent lists all Grafbase API keys, identifies the compromised one by name, deletes it, and confirms the revocation. The operations team completes the rotation through the chat in under a minute, even from their mobile device during off-hours.
A new developer joins and asks what subgraphs exist in production. The AI Agent queries Grafbase for all published subgraphs on the main branch, returns each subgraph name and endpoint URL, and retrieves the federated schema SDL. The developer gets a complete architecture overview from a single conversation instead of hunting through documentation.

Grafbase
FAQs
The agent calls the Get Federated Schema endpoint with an optional branch parameter (defaults to 'main'). It returns the composed GraphQL SDL for the entire federated graph. This is the same schema you would see by running 'grafbase schema' in the CLI, but accessible through conversation.
The agent retrieves schema check results using the Get Schema Check endpoint, which includes pass/fail status, breaking change details, and error summaries. While the agent does not run checks itself (those are triggered via CLI or CI), it can surface results from past checks during conversations.
Tars requires your Grafbase project subdomain and an API key. The subdomain identifies your project for API routing, and the key authenticates requests. Both are entered in the Tars dashboard when configuring the Grafbase tool. Keys can be generated and revoked from your Grafbase project settings.
The API key used for authentication is stored securely in Tars for making requests. Schema data, subgraph information, and audit logs are fetched live from Grafbase during conversations and are not cached or persisted by Tars separately.
Yes. The List Subgraphs endpoint returns all published subgraphs for a specified branch, including each subgraph's name and URL. This gives your team instant visibility into the federation topology without navigating the Grafbase dashboard.
The CLI requires terminal access and knowledge of specific commands. Tars lets anyone on your team query Grafbase through natural language in any chat interface. Non-technical stakeholders can check schema status or audit API access without CLI installation or training.
Yes. The agent can enable or disable MCP servers, list all configured MCP servers, and delete specific server configurations by ID. This makes it possible to manage AI integration points in your Grafbase project through conversational commands.
The API returns an error when a check ID is not found. The agent handles this by informing the user that no matching schema check was located and suggests they verify the check ID or list all available checks to find the correct one. No system errors or broken workflows result.
Don't limit your AI Agent to basic conversations. Watch how to configure and add powerful tools making your agent smarter and more functional.

Privacy & Security
At Tars, we take privacy and security very seriously. We are compliant with GDPR, ISO, SOC 2, and HIPAA.