Turso

@turso which region should I deploy to for lowest latency from my current location?

Based on your client location, the closest Turso region is **iad** (Washington D.C., US East). This region typically offers sub-20ms latency for connections originating from the eastern United States. Deploying your database here will minimize round-trip time for queries. Other nearby options include **ord** (Chicago) and **ewr** (Newark) if you need additional geographic distribution.

This example shows how the MCP uses client geolocation to recommend a deployment region. The tool returns a region code (like 'iad' or 'lhr') based on network proximity. Useful during initial database setup or when planning multi-region replication. Requires a valid Turso API key but does not create or modify any resources.

@turso start listening for changes on the 'orders' table and alert me when new rows are inserted

Now streaming committed changes from the **orders** table. I'll monitor for INSERT, UPDATE, and DELETE events in real-time. When a new order is inserted, I'll surface the row data here in the Space. This stream remains active until you disconnect or the session times out. Note: you'll see changes only after they're committed to the database, not during open transactions.

This demonstrates the real-time change-streaming capability. The MCP establishes a persistent connection to watch a specific table. Useful for building reactive workflows (notify on new signups, trigger actions on status updates). Requires database-level authentication and keeps a connection open—be mindful of concurrent connection limits on your Turso plan.

@turso check if my API token is still valid and when it expires

Your Turso API token is **valid** and active. It expires on **2025-08-14 at 18:32 UTC** (approximately 6 months from now). The token has full read/write permissions for all databases in your organization. You'll need to rotate it before the expiration date to avoid service interruptions.

This example validates token health without making database queries. The MCP confirms the token's active status and returns its expiration timestamp. Helpful for auditing credentials before deploying to production or debugging authentication failures. Does not consume database query quota, but does count against Turso's API rate limits (typically 100 requests/minute).

When Turso MCP makes sense for live deal tracking

A 6-person sales team spread across three continents needs a shared dashboard that updates the second a deal moves stages in their CRM. The Turso MCP's Listen To Changes tool streams table updates without polling, so the AI can surface "Sarah just closed the enterprise deal" in Slack the moment the database commits. The Closest Region tool routes queries to the nearest edge replica, keeping latency under 50ms even for the Sydney rep. This works if your data model is append-heavy (new deals, status changes) and you're comfortable with eventual consistency across regions. If you need strict read-after-write guarantees or complex joins across 10+ tables, a traditional Postgres MCP is safer. Buy this if your team's work is event-driven and geographically distributed.

Turso MCP for high-volume ticket ingestion at scale

A 3-person support team handles 400 tickets a day, and they want the AI to auto-tag incoming issues by scanning a knowledge base stored in Turso. The MCP's Listen To Changes tool watches the tickets table, triggering classification the instant a new row appears. The Closest Region tool ensures the AI queries the replica nearest to the customer's origin, cutting response time by half compared to a single-region database. This setup shines when ticket volume spikes unpredictably (Black Friday, product launches) because Turso scales horizontally without manual intervention. The trade-off: if your knowledge base changes constantly and you need sub-second consistency, the edge replication lag (typically 100-300ms) might surface stale data. Use this if your support load is bursty and your team is tired of provisioning database capacity manually.

When Turso MCP handles per-customer config at the edge

A 5-person product team runs a SaaS with 200 customers, each needing isolated feature flags stored in Turso. The AI uses the Closest Region tool to read flags from the replica nearest to each customer's primary user base, keeping feature checks under 20ms. The Validate API Token tool confirms the MCP's credentials before every session, critical when the AI is making decisions that affect paying customers. This works if your feature flag schema is simple (a few columns, no complex logic) and you're okay with flags taking up to 500ms to propagate globally after a change. If you need instant global consistency or your flags depend on joins across user, account, and billing tables, a single-region Postgres setup is less risky. Buy this if your customers are geographically clustered and you want to avoid the ops overhead of managing per-region databases yourself.