Get agent SQL scope

Error

A valid request URL is required to generate request examples

{
  "results": {
    "schemas": [
      "<string>"
    ],
    "deniedCatalogs": [
      "<string>"
    ],
    "deniedSchemas": [
      "<string>"
    ],
    "catalogs": [
      "<string>"
    ]
  },
  "status": "ok"
}

{
  "error": {
    "name": "<string>",
    "statusCode": 123,
    "data": "<unknown>",
    "message": "<string>"
  },
  "status": "error"
}

GET

api

projects

{projectUuid}

agentSqlScope

Error

A valid request URL is required to generate request examples

{
  "results": {
    "schemas": [
      "<string>"
    ],
    "deniedCatalogs": [
      "<string>"
    ],
    "deniedSchemas": [
      "<string>"
    ],
    "catalogs": [
      "<string>"
    ]
  },
  "status": "ok"
}

{
  "error": {
    "name": "<string>",
    "statusCode": 123,
    "data": "<unknown>",
    "message": "<string>"
  },
  "status": "error"
}

Path Parameters

projectUuid

string<uuid>

required

Stringified UUIDv4. See RFC 4112

Pattern: [0-9A-Fa-f]{8}-[0-9A-Fa-f]{4}-4[0-9A-Fa-f]{3}-[89ABab][0-9A-Fa-f]{3}-[0-9A-Fa-f]{12}

Response

Success

results

object | null

required

Restricts which part of the warehouse the AI agent may read via raw SQL.

Null (or an empty schemas list) means unrestricted — the agent can reach anything the project's warehouse connection can reach, which is the behaviour for every project that has not configured a scope.

This is a correctness control rather than a security boundary: raw SQL already requires manage SqlRunner, so the same data is reachable through the SQL Runner regardless. It exists so an agent can be kept off schemas a customer knows are wrong to answer from, such as a retired dbt project living in the same catalog as the current one.

Show child attributes

status

enum<string>

required

Available options:

ok