TiDB Cloud

@TiDB Cloud list all clusters in our production project and show their current status, region, and tier

Found 4 clusters in project prod-analytics: 1. prod-primary (Dedicated) – Running in us-west-2, 8vCPU/32GB, TiDB v7.5.1 2. prod-replica (Dedicated) – Running in eu-central-1, 4vCPU/16GB, TiDB v7.5.1 3. staging-test (Starter) – Running in us-east-1, shared tier 4. dev-sandbox (Essential) – Paused in ap-southeast-1 All production clusters are healthy. The dev-sandbox cluster has been paused for 14 days to save costs.

This example demonstrates the List Clusters tool's filtering and pagination capabilities. Users need read access to the TiDB Cloud project. The MCP returns one page at a time; for organizations with many clusters, you'll need to handle continuation cursors in follow-up prompts. Cluster details refresh on each call but may lag by 1-2 minutes during scaling operations.

@TiDB Cloud create a new branch called 'feature-user-analytics' from our main cluster for testing schema changes

Branch 'feature-user-analytics' created successfully from cluster main-db (ID: cls_abc123). Branch ID: br_xyz789 Status: Creating (estimated 3-5 minutes) Base snapshot: 2025-01-15 14:32:00 UTC Expiration: 7 days from creation Once active, connect using the branch-specific endpoint. Schema changes in this branch won't affect the parent cluster. Remember to delete the branch when testing is complete to avoid unnecessary charges.

This example shows the branch creation workflow for Starter/Essential instances. Branching requires write permissions and is only available on certain TiDB Cloud tiers. Branches consume storage and compute resources, so they incur costs. The MCP creates the branch but doesn't wait for provisioning to complete—check branch status with a follow-up prompt if you need to confirm it's ready.

@TiDB Cloud pull our January 2025 bill and break down costs by cluster, then identify which cluster had the highest spend

January 2025 TiDB Cloud bill: $4,287.63 total Top 3 clusters by cost: 1. prod-primary (Dedicated): $2,145.20 – 8vCPU tier, 744 hours runtime, $890 storage 2. prod-replica (Dedicated): $1,320.50 – 4vCPU tier, 744 hours runtime, $520 storage 3. analytics-cluster (Dedicated): $680.40 – 4vCPU tier, 372 hours runtime (scaled down mid-month) Remaining $141.53 spread across 6 Starter/Essential instances and data transfer fees. The prod-primary cluster accounts for 50% of total spend due to its larger compute tier and high storage usage.

This example pairs the Get Monthly Bill tool with AI reasoning to surface cost insights. Billing data is read-only and typically available 24-48 hours after month-end. The MCP returns raw line items; the AI synthesizes them into actionable summaries. Users need billing read access at the organization level. Use this for cost attribution, budget tracking, or identifying optimization opportunities.

When TiDB Cloud MCP streamlines finance handoff at month-end

A 6-person engineering team running three TiDB clusters across dev, staging, and prod needs to reconcile cloud spend with finance every month. The Get Monthly Bill tool pulls the full org invoice in one call, while List Clusters maps each line item to the owning project without logging into the console. This MCP is the right call if your team runs more than two clusters and finance asks for a breakdown by environment or cost center. The API key scope covers billing and cluster metadata, so one shared Switchy workspace can handle the monthly ritual without handing out console logins. If your org has fewer than two clusters or you only check spend quarterly, the overhead isn't worth it—just export from the web UI.

When audit log search beats manual console hunting

A 4-person SRE rotation needs to trace who changed a cluster config during a 2am outage. The List Audit Logs tool filters by event type and timestamp, surfacing the exact API calls and user IDs in seconds instead of paging through the console. This MCP wins if your team investigates incidents more than once a quarter or runs compliance reviews that require timestamped change logs. The API key needs audit-read scope, which most orgs already grant to ops leads. The tool returns one page at a time, so if your org generates thousands of events per day, you'll hit pagination limits and need to narrow the time window. For teams that rarely look at logs or have under 50 events per week, the console search is faster than setting up the MCP.

When TiDB Data Service deployments need real-time visibility

A 5-person product team ships a customer-facing API backed by TiDB Data Service and needs to confirm deployments landed before demoing to stakeholders. The List Data App Deployments tool shows live status and error messages without switching contexts from Slack or standup. This MCP is the right call if your team deploys more than twice a week or runs multiple Data Service apps where deployment state isn't obvious from the app itself. The API key requires project-level access, so one workspace can cover all apps in a project. If your team only deploys once a sprint or runs a single app with built-in health checks, the console's deployment view is sufficient and the MCP adds unnecessary indirection.