Tiger MCP

List the PostgreSQL extensions installed in a service database. Connects to a PostgreSQL/TimescaleDB service in Tiger Cloud and returns each installed extension's name, version, schema, and description. The connection is opened in immutable read-only mode.

Execute SQL queries against a service database. Connects to a PostgreSQL database service in Tiger Cloud and executes the provided SQL query, returning the results with column information, row data, and execution metadata. Pass the SQL as 'query', or point 'file' at a .sql file on disk to run its contents. Multi-statement queries (semicolon-separated) are supported when no parameters are provided. All result sets are returned. By default, statements execute in an implicit transaction that automatically commits on success or rolls back on error. Explicit transactions (opened with BEGIN) must be explicitly committed with COMMIT, or they roll back when the connection closes. Process data in the database, not in your context: aggregate, filter, sort/limit, and join in SQL rather than fetching raw rows. WARNING: Can execute any SQL statement including INSERT, UPDATE, DELETE, and DDL commands. Always review queries before execution.

Display the schema of a service database. Connects to a PostgreSQL/TimescaleDB service in Tiger Cloud and returns its schema as readable text: tables (regular, partitioned, and foreign), views, materialized views, enum types, functions, procedures, indexes, triggers, and TimescaleDB hypertable and continuous aggregate metadata. Only objects the connecting role can access are returned. By default only user-facing schemas and objects are shown; view/routine definitions and object comments are omitted unless requested. The connection is opened in immutable read-only mode.

Submit feedback or a bug report to the Tiger Data team. The message reaches a person, so confirm the wording with the user before sending. This opens no support case and returns no ticket to track; for anything that needs a tracked response, point the user at the support URL returned in the output.

Create a new database service in Tiger Cloud with specified type, compute resources, region, and HA options. The default type of service created depends on the user's plan: - Free plan: Creates a service with shared CPU/memory and the 'time-series' add-on - Paid plans: Creates a service with 0.5 CPU / 2 GB memory and the 'time-series' add-on WARNING: Creates billable resources.

Delete a database service permanently. This operation is irreversible: the service and all of its data are destroyed. Deleting a service tagged PROD automatically prompts the user, via an elicitation request through the MCP client, to confirm the deletion before it proceeds, so agents don't need to ask the user for confirmation themselves; if the client cannot prompt, the deletion is refused and the user must run 'tiger service delete' from the CLI instead.

Fork an existing database service to create a new independent copy. You must specify a fork strategy: - 'NOW': Fork at the current database state (creates new snapshot or uses WAL replay) - 'LAST_SNAPSHOT': Fork at the last existing snapshot (faster fork) - 'PITR': Fork at a specific point in time (requires target_time parameter) By default: - Name will be auto-generated from the source service name - CPU and memory will be inherited from the source service - The forked service will be set as the default service WARNING: Creates billable resources.

Get detailed information for a specific database service. Returns connection endpoints, replica configuration, resource allocation, creation time, and status.

List all database services in your Tiger Cloud project. Returns services with status, type, region, and resource allocation.

View logs for a database service. Fetches and displays logs from the specified service. By default, shows the last 100 log entries. Supports filtering by time (via since/until parameters) and node (for services with HA replicas).

List the names of all metric series available for a service, for monitoring, debugging, and performance investigation. Available metrics span hardware/resource usage (CPU, memory, disk, network), PostgreSQL settings, PgBouncer connection-pool stats, PostgreSQL activity/locks/replication, and TimescaleDB internals (compression, chunks, background jobs). Call this first to discover what metrics exist before fetching data with service_metrics_series.

Get descriptive metadata for a metric: what it measures, its type, default aggregation function, and available labels. Use this to understand a metric before querying it for monitoring, debugging, or performance investigation. Use service_metrics_available to discover metric names, this tool to inspect one, then service_metrics_series to fetch its data. These metrics have no richer metadata (returns just the name, with type, default aggregation, description, and labels all empty): timescale_cloud_system_cpu_total_millicores, timescale_cloud_system_cpu_usage_millicores, timescale_cloud_system_disk_io_read_bytes, timescale_cloud_system_disk_io_read_ops, timescale_cloud_system_disk_io_total_bytes, timescale_cloud_system_disk_io_total_ops, timescale_cloud_system_disk_io_write_bytes, timescale_cloud_system_disk_io_write_ops, timescale_cloud_system_disk_usage_bytes, timescale_cloud_system_memory_total_bytes, timescale_cloud_system_memory_usage_bytes.

Fetch time-series data for a named metric over a specified time window, for monitoring, debugging, and performance investigation. Use service_metrics_available first to discover valid metric names. The response groups data points by their label set — a single request may return multiple labeled series (e.g. one per replica, one per worker ordinal). Each series contains its full list of raw data points. Use group_by to control which label keys define that grouping.

Rename a database service. Only the display name changes: the service ID, endpoints, and data are untouched, so existing connections and connection strings keep working.

Resize a database service by changing its CPU and memory allocation. This tool changes the compute resources allocated to your database service. The service may be temporarily unavailable during the resize operation. WARNING: Creates billable resource changes. Increasing resources will increase costs.

Start a stopped database service. This operation starts a service that is currently in a stopped/paused state. The service will transition to a ready state and become available for connections.

Stop a running database service. This operation stops a service that is currently running. The service will transition to a stopped/paused state and will no longer accept connections.

Update the master password for the 'tsdbadmin' user of a database service. The tool generates a secure random password itself; it does not accept one. The change takes effect immediately and may terminate existing connections, and applications using the old password fail to authenticate until they are updated. The new password is saved to the configured password storage and is included in the result only if with_password is true. If the user wants to set a specific password, tell them to run 'tiger service update-password' from the CLI instead, which keeps the password out of this conversation. Updating the password of a service tagged PROD automatically prompts the user, via an elicitation request through the MCP client, to confirm the update before it proceeds, so agents don't need to ask the user for confirmation themselves; if the client cannot prompt, the update is refused and the user must run 'tiger service update-password' from the CLI instead.