Provider REST API

8 actions

Connect Claude to Snowflake

Browse databases, schemas and tables, and run SQL through the Snowflake SQL API. Toolspoke puts 8 of its actions behind one MCP endpoint that Claude, Cursor, and Codex all speak.

Connect View pricing

Connection: Provider REST API
Authentication: Programmatic access token
Actions exposed: 8
Cost per call (typical): 1 credit
Adapter: Maintained by Toolspoke

Connected in three steps

1
Install Snowflake
Open the marketplace in your workspace, add Snowflake to the project your agents work in, and it appears on the gateway immediately.
2
Connect the credential
Authenticate with programmatic access token. Where to get one, and what it has to be able to reach, is the next section.
3
Point your agent at the gateway
Give your client one address, https://toolspoke.com/mcp. Claude Code takes it as a command, Claude and Claude Desktop add it as a custom connector, and Cursor, Codex and VS Code each read it from a config file of their own.

.mcp.json

{
  "mcpServers": {
    "toolspoke": {
      "type": "http",
      "url": "https://toolspoke.com/mcp"
    }
  }
}

One block covers every tool you have installed. Snowflake shows up in the client as soon as your policy allows it, and so does everything else you install later.

Where the address goes, per client

Claude Code: Run it in your project, then /mcp to sign in
claude mcp add --transport http toolspoke https://toolspoke.com/mcp
Claude and Claude Desktop: Settings, then Connectors, then Add custom connector
https://toolspoke.com/mcp
Cursor: ~/.cursor/mcp.json, or .cursor/mcp.json for one project
{ "mcpServers": { "toolspoke": { "url": "https://toolspoke.com/mcp" } } }
Codex: ~/.codex/config.toml
[mcp_servers.toolspoke] url = "https://toolspoke.com/mcp"
VS Code: .vscode/mcp.json, or the MCP: Add Server command
{ "servers": { "toolspoke": { "type": "http", "url": "https://toolspoke.com/mcp" } } }

What Snowflake asks for

Programmatic access token. You provide it once, when you install the connector. Toolspoke encrypts it at rest and decrypts it only for the length of a single call, and the gateway attaches it to the outbound request itself, so it is never part of the arguments an agent sends.

Account identifierRequired: The organization-account name from Snowsight → Account details, written with a hyphen. It becomes the host: https://myorg-myaccount.snowflakecomputing.com.; myorg-myaccount
Programmatic access tokenRequired: Snowsight → your user → Programmatic access tokens, or ALTER USER <name> ADD PROGRAMMATIC ACCESS TOKEN <token_name>. The secret is shown once. It inherits the role you bind it to - bind it to a read-only role unless this install must write.
Default warehouseOptional: Warehouse used by submit_statement when a call does not name one. Without a warehouse Snowflake cannot run a query at all.; COMPUTE_WH
Default roleOptional: Role used when a call does not name one. Must be a role the token's user already has.; ANALYST_RO
Default databaseOptional: Database context for statements that use unqualified table names.; ANALYTICS
Default schemaOptional: Schema context for statements that use unqualified table names.; PUBLIC

What Claude can do in Snowflake

8 actions, each one declared and named by the connector rather than discovered at runtime. A workspace policy grants a person all of them, a hand-picked selection, everything on the read side, everything on the write side, or none.

Reads: 6Reads
Writes: 1Writes
Destructive: 1Destructive

Reads

Fetches data and changes nothing.

list_databases
List the databases the token's role can see, with owner, comment, retention time and creation date. Call this first - every schema, table and describe operation needs a database name, and this is the cheapest call in the connector: it runs no warehouse and costs no credits.
list_schemas
List the schemas inside one database. Use it between list_databases and list_tables to find where the tables you want actually live - a large Snowflake account puts raw, staging and modelled data in different schemas of the same database. Costs no credits.
list_tables
List the tables in one schema, with row counts and bytes where Snowflake tracks them. Use it to find a table name before describe_table, and to judge how large a table is before writing a query against it. Costs no credits.
describe_table
Describe one table: every column with its type, nullability, default and comment, plus the table's clustering key and comment. Call this before writing any statement against a table you have not seen - it is what stops a query failing on a column that does not exist, and it costs no credits, unlike discovering the same thing by running a SELECT.
list_warehouses
List the virtual warehouses the token's role can use, with size, state and auto-suspend settings. Call this when submit_statement reports no warehouse, or to pick the smallest warehouse that can answer a question - the warehouse is what a query costs credits on. Costs no credits itself.
get_statement
Fetch the status and results of a statement already submitted, by its statementHandle. A 202 means it is still running - poll again; a 200 carries `resultSetMetaData` and `data`. This is how you finish an async submit_statement, and how you retrieve a further page of a large result set by passing `partition`.

Writes

Creates or updates something on the other side.

submit_statement
Run one SQL statement on a warehouse and return its result set. Classified as a write because Snowflake's SQL API has no read-only mode: the same endpoint serves SELECT, INSERT and DDL, and nothing here can prove which one a caller sent. Bind the access token to a read-only role if this install should only ever read. Always put an explicit LIMIT on a SELECT - the rows come back verbatim and a wide table will exhaust your context. Use the metadata reads (list_tables, describe_table) for schema questions instead; they cost no credits, and this does. If the response has no `data`, the statement is still running: take `statementHandle` and poll get_statement.

Destructive

Deletes or permanently alters something. Worth granting on purpose.

cancel_statement
Cancel a statement that is still running, by its statementHandle. Destructive because it aborts work in progress: a long INSERT or MERGE stops part-way and Snowflake rolls it back, and the credits already burned on the warehouse are not returned. Use it on a runaway query found through get_statement, not as a routine cleanup step.

What it will not do

Enforced by the gateway rather than left to convention, which is why each of these can be stated flatly.

It cannot call anything else: The 8 actions above are the whole of it. A call to any other name is refused before it reaches Snowflake rather than forwarded on, and connecting your account does not add to the list: it is fixed by the connector, not discovered at run time.
It reaches no further than your credential: Toolspoke holds no access to Snowflake of its own. Every call carries the credential you stored and nothing besides, so whatever that credential cannot reach, this connector cannot reach either.
It never hears from Snowflake: Nothing is pushed to it. There is no webhook, no subscription and no polling, so this connector cannot notice by itself that something changed in Snowflake. An agent has to ask.
It does not smooth over provider limits: Toolspoke does not retry, queue or back off around Snowflake's own rate limits. A call that Snowflake refuses comes back to the agent as a failed call.

Before you connect it

What can Claude do in Snowflake?

8 named actions: 6 that only read, 1 that write and 1 that delete or permanently alter something. They include list_databases, list_schemas and list_tables. Nothing outside that list is reachable: the connector declares each operation by name rather than proxying whatever an agent asks for.

What credentials does the Snowflake connector need?

Programmatic access token. The connector asks for account identifier and programmatic access token, and optionally default warehouse, default role, default database and default schema. Values are encrypted at rest and attached to the outbound request by the gateway, so they are never part of the arguments an agent sends and never reach the audit log.

Does the Snowflake connector work with Cursor and Codex, or only Claude?

Any client that speaks MCP, and every one of them gets the same 8 actions. There is a single address, https://toolspoke.com/mcp. Claude Code adds it with claude mcp add --transport http, Claude and Claude Desktop take it as a custom connector in settings, Cursor reads it from .cursor/mcp.json, Codex from ~/.codex/config.toml, and VS Code from .vscode/mcp.json. Each of them signs in to the gateway itself, so there is no key to paste.

What does the Snowflake connector not do?

The 8 actions above are the whole of it. A call to any other name is refused before it reaches Snowflake rather than forwarded on, and connecting your account does not add to the list: it is fixed by the connector, not discovered at run time. Toolspoke holds no access to Snowflake of its own. Every call carries the credential you stored and nothing besides, so whatever that credential cannot reach, this connector cannot reach either. Nothing is pushed to it. There is no webhook, no subscription and no polling, so this connector cannot notice by itself that something changed in Snowflake. An agent has to ask. Toolspoke does not retry, queue or back off around Snowflake's own rate limits. A call that Snowflake refuses comes back to the agent as a failed call.

Can I limit which actions an agent can call?

Yes, in two places. The project switches Snowflake's actions on and off one at a time, for everyone in the project at once, and the screen groups them by read, write and destructive so turning off everything that deletes is one click. An individual agent key can then be narrowed further, to particular toolkits in a project and to particular actions in a toolkit. Whatever it was granted, a key never reaches a project its owner cannot.

What gets recorded when an agent calls Snowflake?

Every attempt, with the agent that made it and the person that agent belongs to, the full request payload, the response payload, the status, the duration, and the credits spent. Values whose key names a secret are masked out before the record is shown to anyone. An operation the connector marks as not retained never has its response body written at all, so the gateway keeps no second copy of what was read.

More data connectors

Browse all of them