Retrieve every LLM provider the AI agents of a pipe can use: Pipefy-managed (system) providers and the custom providers of the pipe organization.
Before You Begin
🔗 Use the GraphQL Playground to execute the queries in this guide.
➡️ New to GraphQL? Learn how to navigate the Playground with our Playground Basics Guide.
Prerequisites
- Authentication: Use a Service Account token.
- Permissions: You must be able to
manage AI agentsin the pipe, which is granted to the pipeadminrole. - Pipe UUID: The UUID of the pipe whose AI agents will use the providers.
Step 1: Execute the Query
This query returns the LLM providers available when configuring the AI agents of a pipe: the system providers (managed by Pipefy) and the custom providers configured by the organization that owns the pipe. Unlike List All LLM Providers by Organization, it only requires permission on the pipe, not on the organization's AI provider settings.
Each node is an LlmProviderUnion. Use inline fragments (... on LlmProvider / ... on SystemLlmProvider) to select the fields of each type, and __typename to tell them apart.
query {
allLlmProvidersByPipe(pipeUuid: "pipe-uuid-example", onlyActiveProviders: true) {
edges {
cursor
node {
__typename
... on LlmProvider {
id
name
configuration
active
organizationDefault
}
... on SystemLlmProvider {
id
name
configuration
state
deprecationDate
organizationDefault
}
}
}
pageInfo {
hasNextPage
endCursor
}
}
}
Arguments
pipeUuid(ID, required): The UUID of the pipe.onlyActiveProviders(Boolean, optional, defaultfalse): Whentrue, filters out disabled custom providers. System providers are always returned.
Response Fields
The response has the same shape as List All LLM Providers by Organization:
allLlmProvidersByPipe: A connection object whose nodes are anLlmProviderUnion.edges.node: Either aLlmProvider(custom) or aSystemLlmProvider(Pipefy-managed), identified by__typename.pageInfo: Pagination details (hasNextPage,endCursor).
Example Response
{
"data": {
"allLlmProvidersByPipe": {
"edges": [
{
"cursor": "MQ",
"node": {
"__typename": "SystemLlmProvider",
"id": "1",
"name": "Pipefy AI",
"configuration": {
"provider": "pipefyai",
"model": "gpt-4o"
},
"state": "active",
"deprecationDate": null,
"organizationDefault": true
}
},
{
"cursor": "Mg",
"node": {
"__typename": "LlmProvider",
"id": "27",
"name": "OpenAI GPT-4o",
"configuration": {
"provider": "openai",
"model_name": "gpt-4o"
},
"active": true,
"organizationDefault": false
}
}
],
"pageInfo": {
"hasNextPage": false,
"endCursor": "Mg"
}
}
}
}
Key Notes
- The providers listed are those of the organization that owns the pipe.
- The query returns a
RESOURCE_NOT_FOUNDerror when no pipe matchespipeUuid(including the UUID of a database table), andPERMISSION_DENIEDwhen you cannot manage the pipe's AI agents (pipe members, for example). - Sensitive information like API keys is never exposed in the
configurationobject.

