Connecting AI Tools
Step-by-step setup for connecting Cursor, Claude, and ChatGPT to Seemore over MCP, plus personal access tokens for scripts and the REST API.
Last updated
Step-by-step setup for connecting Cursor, Claude, and ChatGPT to Seemore over MCP, plus personal access tokens for scripts and the REST API.
This page walks through connecting an AI assistant to Seemore over MCP, and through creating a personal access token for scripts and the REST API.
Start by opening Settings → API Access in Seemore. You'll find your personalized MCP URL there — copy it before you begin.
AI tools sign in through your browser — there is no token to paste. After you add the URL, the tool opens a Seemore sign-in tab, you approve access, and the connection is ready. Approved tools appear under Connected AI tools.
Open Settings → Cursor MCP and add this block to your MCP configuration, replacing the URL with the one from Settings → API Access:
{
"mcpServers": {
"seemore": {
"url": "<your-mcp-url>",
"type": "streamable-http"
}
}
}Cursor will prompt you to sign in to Seemore in your browser the first time it connects.
In Claude, go to Settings → Connectors → Add custom connector, give it a name (for example, Seemore), and paste your MCP URL:
<your-mcp-url>Claude opens a browser tab to sign you in to Seemore and complete the connection.
In the ChatGPT web app, go to Settings → Apps, turn on Developer mode, click Create app, and paste your MCP URL:
<your-mcp-url>ChatGPT opens a browser tab to sign you in to Seemore and complete the connection.
Any MCP-compatible client that supports remote servers over Streamable HTTP can connect using the same URL and browser sign-in.
Once connected, try prompts like:
"Summarize my Snowflake spend over the last 30 days, broken down by warehouse."
"What's driving my biggest cost anomaly this week? Walk me through the root cause."
"Show the lineage downstream of ANALYTICS.PUBLIC.ORDERS and tell me which tables nobody reads."
"Find queries with remote spillage in the last 7 days and recommend fixes."
"List my waste-reduction insights sorted by annual savings."
For scripts, partner integrations, and curl, create a personal access token and send it as a Bearer token. Unlike AI tools, these do not use browser sign-in.
To create a token:
Go to Settings → API Access → Tokens and connected tools.
On the Personal access tokens tab, enter a descriptive name (for example, ETL pipeline) and click Create token.
Copy the token immediately — the plaintext is shown only once. Store it in your script's secret store.
Treat personal access tokens like passwords. A token carries your access; anyone holding it can read everything your role can. Revoke a token from the same screen if it is ever exposed — it stops working immediately on its next call.
Set your token as an environment variable and call the REST endpoints. The REST base is https://external-api.seemoredata.io/v1.
The full request and response schema for every operation is published as OpenAPI at https://external-api.seemoredata.io/swagger.
All connected tools and tokens are listed under Settings → API Access → Tokens and connected tools:
Personal access tokens — shows each token's name, when it was created, and when it was last used. Revoke a token to invalidate it immediately.
Connected AI tools — shows each AI client that signed in, when it connected, and when it was last used. Revoking a tool invalidates its access; the tool must sign in again to reconnect.
Last updated
# Inspect available operations
curl -H "Authorization: Bearer $SEEMORE_TOKEN" \
https://external-api.seemoredata.io/v1
# Search assets
curl -H "Authorization: Bearer $SEEMORE_TOKEN" \
-H "Content-Type: application/json" \
-d '{"query":"orders","limit":25}' \
https://external-api.seemoredata.io/v1/search-assets