Tiger MCP reference
Tool reference for Tiger MCP, the Model Context Protocol server bundled with Tiger CLI
Tiger MCP gives your AI agent access to Tiger Cloud so you can manage services and query your data in natural language. It is built into the Tiger CLI binary.
To install and configure Tiger MCP for your AI agent, see Integrate Tiger Cloud with your AI agent. This page is the reference for the Tiger MCP tools your agent can call. For an overview of Tiger MCP and Tiger CLI, see Tiger CLI and Tiger MCP.
Tiger MCP tools
Section titled “Tiger MCP tools”Tiger MCP exposes the following tools to your AI agent. You do not call these directly; you describe what you want and the agent selects the tool.
Parameter names, types, and required or optional fields can change with new Tiger MCP versions. Run tiger mcp get <tool_name> for the current definition of any tool.
Service tools
Read-only mode restricts the mutating tools: service_create, service_fork, service_start, service_stop, service_resize, service_rename, service_update_password, service_delete. With read_only=all they are not registered at all; with read_only=prod they stay available but refuse calls against services tagged PROD (for service_create and service_fork, a call that requests environment: PROD). See Restrict Tiger MCP to read-only.
service_list
List the services in the current project. No parameters.
Returns: an array of services, each with its status, type, region, and resource allocation.
service_get
Show detailed information about one service.
| Parameter | Type | Required | Description |
|---|---|---|---|
service_id | string | Yes | The target service. |
with_password | boolean | No | Include the password. Only set this if you explicitly ask for the password. Default: false. |
Returns: connection endpoints, replica configuration, resource allocation, creation time, and status.
service_create
Create a new service.
| Parameter | Type | Required | Description |
|---|---|---|---|
name | string | No | Service name (auto-generated if omitted). |
addons | array of strings (time-series) | No | Addons to enable; time-series enables TimescaleDB. Use an empty array for PostgreSQL-only. Before Tiger CLI 0.26, there was also a deprecated ai addon. |
region | string | No | Cloud region, for example us-east-1. |
cpu_memory | string (enum) | No | CPU/memory allocation, for example "4 CPU/16 GB". See allowed configurations. |
replicas | integer (0-5) | No | Number of high-availability replicas. Default: 0. |
environment | string (enum: DEV, PROD) | No | Environment tag for the new service. Default: DEV. Under read_only=prod, requesting PROD is refused. |
wait | boolean | No | Wait for the service to be ready before returning. Default: false. |
set_default | boolean | No | Set the new service as the default. Default: true. |
with_password | boolean | No | Include the password in the response. Default: false. |
Returns: the new service's details, including its ID, connection endpoint, and status.
service_create provisions billable infrastructure.
service_fork
Fork a service into an independent copy.
| Parameter | Type | Required | Description |
|---|---|---|---|
service_id | string | Yes | The source service. |
fork_strategy | string (enum: NOW, LAST_SNAPSHOT, PITR) | Yes | When to fork from. |
target_time | string (RFC3339) | Only with PITR | Point in time to fork from. |
name | string | No | Fork name (auto-generated if omitted). |
cpu_memory | string (enum) | No | CPU/memory allocation for the fork. Inherits from the source if omitted. |
environment | string (enum: DEV, PROD) | No | Environment tag for the fork. Default: DEV. Under read_only=prod, requesting PROD is refused; forking a PROD source into a DEV fork is allowed. |
wait | boolean | No | Wait for the fork to be ready before returning. Default: false. |
set_default | boolean | No | Set the fork as the default service. Default: true. |
with_password | boolean | No | Include the password in the response. Default: false. |
Returns: the new fork's details, including its ID, connection endpoint, and status.
service_fork provisions billable infrastructure.
service_resize
Change a service's CPU and memory.
| Parameter | Type | Required | Description |
|---|---|---|---|
service_id | string | Yes | The service to resize. |
cpu_memory | string (enum) | Yes | New CPU/memory allocation, for example "4 CPU/16 GB". |
wait | boolean | No | Wait for the resize to finish before returning. Default: false. |
Returns: the service's updated resource allocation and status.
service_resize affects billing, and the service may be briefly unavailable.
service_start
Start a stopped service.
| Parameter | Type | Required | Description |
|---|---|---|---|
service_id | string | Yes | The service to start. |
wait | boolean | No | Wait for the service to be fully started before returning. Default: false. |
Returns: the service's updated status.
service_stop
Stop a running service.
| Parameter | Type | Required | Description |
|---|---|---|---|
service_id | string | Yes | The service to stop. |
wait | boolean | No | Wait for the service to be fully stopped before returning. Default: false. |
Returns: the service's updated status.
service_rename
Rename a service. Only the display name changes: the service ID, endpoints, and stored credentials are untouched. The CLI equivalent is tiger service rename.
| Parameter | Type | Required | Description |
|---|---|---|---|
service_id | string | Yes | The service to rename. |
name | string | Yes | The new name for the service. |
Returns: the service ID (unchanged) and the name as stored after the rename.
service_update_password
Update the tsdbadmin password. The tool generates a secure random password itself; it does not accept one, because a password passed as a parameter, or typed into the chat, ends up in the agent's context. To set a specific password, use tiger service update-password from the CLI instead, which keeps it out of the conversation. Before Tiger CLI 0.26, this tool instead took a required password parameter.
The change takes effect immediately and may disconnect existing sessions; applications using the old password fail to authenticate until they are updated. The new password is saved to the configured password storage. If saving fails, or password storage is off, and with_password is false, the result warns you that nobody has the password.
| Parameter | Type | Required | Description |
|---|---|---|---|
service_id | string | Yes | The service to update. |
with_password | boolean | No | Include the generated password in the response. Default: false. |
Returns: whether the password was updated, where it was saved locally, and the password itself when with_password is true. updated is false when you decline the confirmation prompt.
When the target is tagged PROD, the tool prompts you through the MCP client to type the service ID back before it proceeds, like service_delete. If your MCP client doesn't support this prompt (MCP elicitation), the update is refused and you are pointed to tiger service update-password instead.
service_delete
Delete a service permanently. The CLI equivalent is tiger service delete. Before Tiger CLI 0.25, there was no Tiger MCP tool for delete.
| Parameter | Type | Required | Description |
|---|---|---|---|
service_id | string | Yes | The service to delete. |
Returns: the service ID, whether it was deleted, and a human-readable outcome. deleted is false when you decline the confirmation prompt.
Deleting a service destroys it and all of its data. When the target is tagged PROD, the tool prompts you through the MCP client to type the service ID back before it proceeds, matching the CLI's confirmation. If your MCP client doesn't support this prompt (MCP elicitation), the deletion is refused and you are pointed to tiger service delete instead, so a production service is never deleted unconfirmed.
service_logs
Fetch service logs.
| Parameter | Type | Required | Description |
|---|---|---|---|
service_id | string | Yes | The service to fetch logs for. |
tail | integer | No | Number of log lines to return. Default: 100. |
since, until | string (RFC3339) | No | Restrict logs to a time range. |
node | integer | No | Specific node to fetch logs from (for services with HA replicas; 0 is valid). Defaults to the primary node. |
Returns: the matching log lines.
Metrics tools
These tools let your agent monitor, debug, and investigate the performance of a service from its time-series metrics. They are read-only, so they stay available in read-only mode. The CLI equivalents are the tiger service metrics commands; see Available metrics for the full list of metrics with descriptions. For a worked debugging workflow, see Diagnose a slowdown with service metrics. Before Tiger CLI 0.26, the metrics tools were experimental and only registered when the TIGER_EXPERIMENTAL environment variable was set.
service_metrics_available
List the names of the metric series available for the given service.
| Parameter | Type | Required | Description |
|---|---|---|---|
service_id | string | Yes | The target service. |
Returns: an array of metric series names, for use with service_metrics_details and service_metrics_series.
service_metrics_details
Get descriptive metadata for a metric: what it measures, its type, its default aggregation function, and the labels it can be filtered or grouped by. The timescale_cloud_system_* metrics have no richer metadata; for those, the tool returns just the name, with the other fields empty.
| Parameter | Type | Required | Description |
|---|---|---|---|
service_id | string | Yes | The target service. |
metric_name | string | Yes | The metric to describe. Use service_metrics_available to discover valid names. |
Returns: the metric's name, type, default aggregation, description, and labels.
service_metrics_series
Fetch time-series data for a metric over a time window. The response groups data points by their label set, so a single request can return multiple labeled series, for example one per replica. Before Tiger CLI 0.26, filters supported only equality matches.
| Parameter | Type | Required | Description |
|---|---|---|---|
service_id | string | Yes | The target service. |
metric_name | string | Yes | The metric series to fetch. Use service_metrics_available to discover valid names. |
from, to | string (RFC3339) | Yes | The time window. to cannot be in the future; use the current time for the latest data. |
role | string (enum: PRIMARY, REPLICA) | No | Convenience filter for the role label. Omit to aggregate datapoints across all nodes. For most queries applying PRIMARY could be more meaningful. |
filters | array of {key, value, match_type} | No | Arbitrary label filters. match_type is EQUAL (default) or NOT_EQUAL; NOT_EQUAL also matches series missing the label entirely. |
bucket_seconds | integer | No | Aggregation bucket size in seconds (minimum: 60). When omitted, the server picks a size matched to the window. |
fn | string (enum) | No | Aggregation function applied per bucket: RATE, INCREASE, SUM, AVG, MIN, MAX, MIN_TOTAL, MAX_TOTAL, COUNT, P50, P90, P99, or LAST. Omit to let the server pick the metric's default. Not accepted on the timescale_cloud_system_* metrics. See Aggregation functions for what each one computes. |
group_by | array of strings | No | Label keys to break the result into one series per distinct value combination. Rejected on the same metrics that reject fn. Omit to collapse into a single series. |
Returns: one series per distinct label set, each with its full list of data points.
Database tools
db_query
Run SQL against a service. The CLI equivalent is tiger db query. Before Tiger CLI 0.24, this tool was named db_execute_query.
| Parameter | Type | Required | Description |
|---|---|---|---|
service_id | string | Yes | The target service (or a read replica set ID, to query the replica instead of the primary). |
query | string | Exactly one of query or file | The SQL to run. |
file | string | Exactly one of query or file | Path to a SQL file on disk to run, instead of passing its contents as query. |
parameters | array of strings | No | Values substituted for $1, $2, and so on. Not supported with multi-statement queries. |
timeout_seconds | integer | No | Query timeout. Default: 30. |
role | string | No | Database role to connect as. Default: tsdbadmin. |
pooled | boolean | No | Use connection pooling. Default: false. |
Returns: column information, row data, and execution metadata for each statement. Values come back exactly as PostgreSQL's text format renders them, the same text psql shows, with SQL NULL as null. To limit the data sent to the agent, a result set is truncated when it exceeds the mcp_max_rows configuration option (default: 100) or an overall response-size cap; the response then flags the truncation and the agent is told to refine the query in SQL instead.
db_query can run INSERT, UPDATE, DELETE, and DDL. Multi-statement queries (semicolon-separated) are supported when no parameters are provided. In read-only mode, writes and DDL are rejected by the server.
db_schema
Return the schema of a service database as readable text (tables, views, materialized views, and more).
| Parameter | Type | Required | Description |
|---|---|---|---|
service_id | string | Yes | The target service (or a read replica set ID). |
schema | string | No | Restrict output to a single schema. |
definitions | boolean | No | Include full object definitions. Default: false. |
comments | boolean | No | Include object comments (COMMENT ON text). Default: false. |
internal | boolean | No | Include system schemas and extension-owned objects. Default: false. |
role | string | No | Database role to connect as. Default: tsdbadmin. |
pooled | boolean | No | Use connection pooling. Default: false. |
Returns: readable text describing tables, views, materialized views, enum types, functions, procedures, indexes, triggers, and TimescaleDB hypertable and continuous aggregate metadata. The connection is opened in immutable read-only mode, so this tool never writes.
db_extensions_list
List the PostgreSQL extensions installed in a service database, so your agent can check what is available, such as timescaledb_toolkit, before it writes SQL that depends on it. The CLI equivalent is tiger db extensions list. Added in Tiger CLI 0.27.
| Parameter | Type | Required | Description |
|---|---|---|---|
service_id | string | Yes | The target service (or a read replica set ID, to list the extensions on the replica instead of the primary). |
role | string | No | Database role to connect as. Default: tsdbadmin. |
Returns: the installed extensions, ordered by name, each with its name (as used in CREATE EXTENSION), installed version, the schema its objects live in, and its description, when it has one. The connection is opened in immutable read-only mode, so this tool never writes and stays available in read-only mode.
Documentation and skills tools
search_docs
Search Tiger Data documentation with hybrid semantic (vector) and keyword search.
| Parameter | Type | Required | Description |
|---|---|---|---|
source | string (enum) | Yes | The documentation source, for example tiger, postgres_17, or postgis_3.5. |
query | string | Yes | The search query. |
limit | integer | Yes | Maximum matches to return. Default: 20. |
semanticWeight | number (0-1) | Yes | 0 for keyword-only, 1 for semantic-only, or a value in between to blend the two. Default: 0.7. |
Returns: ranked matches from the requested documentation source.
view_skill
Retrieve a built-in skill for TimescaleDB operations and best practices (for example, schema design, hypertable setup, and migration planning).
| Parameter | Type | Required | Description |
|---|---|---|---|
skill_name | string | Yes | The skill to retrieve, or . to list all available skills. |
path | string | Yes | A relative path to a file or directory within the skill. Use . to list the skill's root directory, or leave empty for the skill's SKILL.md. |
Returns: the requested skill content, or a directory listing.
If your coding agent supports skills, you can also install the same skills locally with tiger skills install, so the agent loads them natively instead of calling view_skill.
Feedback tool
feedback
Send feedback or a bug report to the Tiger Data team. The message reaches a person, so your agent should confirm the wording with you before sending. It opens no support case and returns no ticket; the response includes the Tiger Console URL for opening a tracked support ticket. The CLI equivalent is tiger feedback.
| Parameter | Type | Required | Description |
|---|---|---|---|
message | string | Yes | The feedback or bug report: what you were trying to do and what happened. At most 3,000 characters. |
Returns: whether the message was sent, and the support URL for a tracked ticket.
Manage the Tiger MCP server
Section titled “Manage the Tiger MCP server”To install, list, inspect, and start Tiger MCP, use the tiger mcp commands, documented in the Tiger CLI reference. Global flags that apply when running Tiger MCP are in the configuration parameters.
Tiger MCP ships inside the open-source Tiger CLI. File an issue in the repo to report bugs or request features and help shape the product.