How is this different from connecting MotherDuck MCP directly?

Connected directly, an agent sees every tool the server exposes and holds the credential itself. Through TrustGate you choose which tools each agent gets, and every call is logged against the agent that made it.

What if the MotherDuck MCP API changes?

We track the MotherDuck MCP schema, so the tool catalogue stays current. Tools you have already allowed keep working, and anything new stays denied until you enable it.

Can I give different agents different MotherDuck tools?

Yes. Access is set per agent: one can get a handful of read tools, another the full set. You grant the whole MCP or single tools within it.

Where can I see what my agents did in MotherDuck?

Every call is logged against the agent that made it, with the tool it used. Activity can be sent on to your own monitoring.

Do I need to run the MotherDuck MCP server myself?

No. TrustGate connects to the MotherDuck MCP server and sits in front of it. Your agent only talks to your TrustGate URL, never to MotherDuck directly.

Connect MotherDuck MCP to your AI agents, securely.

Q: Is MotherDuck MCP secure?

The server itself exposes 5 tools, and some of them can write to your MotherDuck account. Behind TrustGate an agent only sees the tools you allow, requests and responses are inspected, and every call is logged. Anything outside the allowlist is denied before it reaches MotherDuck.

One secured endpoint for every MotherDuck call your AI agents make, scoped per user, inspected, and logged.

Get started for free

Get a demo

Try MotherDuck MCP now

What your agents can ask once MotherDuck MCP is connected.

Select the tools your AI agents can call

5 tools from the MotherDuck MCP schema.

Showing 5 tools

list_databasesreadLists all databases, which is useful for MotherDuck or when multiple databases are attached.
list_tablesreadLists tables and views, optionally scoped to a database and schema.
list_columnsreadLists the columns of a table or view, optionally scoped to a database and schema.
execute_queryhigh riskExecutes an arbitrary SQL query in the DuckDB dialect. Results return as JSON capped at 1024 rows and 50,000 characters by default. Write access requires the server to be started with --read-write.
switch_database_connectionhigh riskSwitches the server to a different database by path, which can be a local file, an S3 URL or MotherDuck, optionally creating it if it does not exist. It is only exposed when the server is started with --allow-switch-databases.

What MotherDuck MCP exposes

MotherDuck MCP groups its 5 tools into 3 categories, each with a suggested starting policy.

All 5 tools as JSON, ready to paste.

Category	Covers	Access	Suggested policy
Catalog	Listing all databases, which matters for MotherDuck or multiple attached databases, listing the tables and views in a database or schema, and listing the columns of a table or view	Read	Read-only and the right prerequisite for a query that compiles. These three are cheap to allow freely and keep an agent from guessing at schema it cannot see
SQL Execution	Executing an arbitrary SQL query in the DuckDB dialect against whichever database the connection currently points at, returning JSON capped at 1024 rows and 50,000 characters by default	Write	DuckDB reaches past the attached database to the local filesystem and its own settings, which is why MotherDuck says read-only mode is not sufficient on its own. Log the SQL text rather than the tool name, and set a query timeout, since the default is disabled
Connection	Switching the server to a different database by path, optionally creating it if it does not exist, where the path can be a local file, an S3 URL or MotherDuck	Write	This is hidden unless --allow-switch-databases is passed, which is the cleanest control here. Leaving it off pins the connection to one database for the life of the server

Connect MotherDuck MCP to your AI agent

Four steps, under ten minutes.

01
Sign in to NeuralTrust
Open the NeuralTrust platform, sign in and go to Getting started, then Connect your tools.
02
Choose your agent
When asked which agent you use, pick Claude Code, Codex, Cursor, Windsurf, Copilot or your own code.
03
Connect MotherDuck
Find MotherDuck in the tool list and connect it.
04
Connect your agent
Follow the steps on screen to add the config to your agent, then send the test prompt.

Permissions down to the single tool

You do not allow MotherDuck MCP or block it. You allow the exact tools each agent can call, one by one.

MotherDuck MCP	5 TOOLS AVAILABLE
Code review agent	4 tools, read only
Issue triage bot	9 tools
Platform team assistant	all 5 tools

Tool by toolConnecting the server is not the unit of access. Each tool is granted on its own.
Default denyA tool you have not enabled is never exposed to the agent.
Approval on writesTools that change or delete things can require a human to confirm.

Is MotherDuck MCP secure?

One endpoint instead of raw credentials. Every call is checked against policy and logged.

AI agentsConnect through your MCP client configuration.	NeuralTrustApply controls based on your configuration.	MotherDuck MCP5 documented tools in the local DuckDB and MotherDuck MCP server, run through uvx over stdio or as a Docker container over streamable HTTP. It defaults to an in-memory database in read-only mode; --read-write enables writes and --allow-switch-databases exposes the fifth tool. Query timeout is disabled by default, and MotherDuck states that read-only mode alone is not sufficient for third-party access.
Coding assistants	Connection & credentialsper user	Catalog3 tools
Internal copilots	Tool access policiesdeny by default	SQL Execution1 tool
Workflow automations	Request & response inspectioninline	Connection1 tool
Compatible MCP clients	Activity monitoringalerts and logs