Using it from n8n

There are two ways to send a workflow's calls to the mock.

	Base-URL mode	Proxy mode
How	Change the URL a node calls	Point all of n8n's traffic at the mock
Works for	HTTP Request nodes, and nodes whose credential has a base URL	Every node, including Slack, Linear, Airtable and HubSpot
Setup	None	Three environment variables, or `integration-mock up`
n8n Cloud	Yes	No

The same pack gives the same answers in both modes.

Base-URL mode

integration-mock start
integration-mock packs enable slack
integration-mock url slack

This prints http://127.0.0.1:8080/slack. Set the HTTP Request node's URL to that address plus the path, for example http://127.0.0.1:8080/slack/api/chat.postMessage.

An n8n HTTP Request node pointed at the mock, with the mocked response in the output panel

The same workflow after Execute workflow: both nodes green

Which nodes can be pointed at the mock

Node	Base-URL mode
HTTP Request	Yes
A node whose credential has a base URL, such as Salesforce, Gong or Supabase	Yes, through the credential
A node with a fixed URL, such as Slack, Linear, Airtable, HubSpot or Google Sheets	No. Use proxy mode

Where n8n runs

The mock only needs an address n8n can reach.

n8n runs	Address to use
On your machine	`http://127.0.0.1:8080/<service>`
In Docker, same network as the mock	`http://mock:8080/<service>`
On a server	The mock's private address on that network
n8n Cloud	A public address for the mock, such as a tunnel or a small server

Warning

Never expose the admin port, 8081. The mock port itself has no authentication either, so put it behind something you control.

The mock serves plain HTTP. Put a reverse proxy or a tunnel in front of it if you need HTTPS.

Proxy mode

In proxy mode, n8n sends every outgoing call through the mock. Calls to services you enabled are answered by the mock. Everything else passes straight through, untouched and unlogged.

With Docker

integration-mock up      # n8n on http://localhost:5690, already pointed at the mock
integration-mock down

With your own n8n

Run integration-mock ca install to print the variables, set them where n8n runs, and restart n8n.

HTTP_PROXY=http://127.0.0.1:8080
HTTPS_PROXY=http://127.0.0.1:8080
NODE_EXTRA_CA_CERTS=~/.integration-mock/ca.pem
NO_PROXY=localhost,127.0.0.1

NODE_EXTRA_CA_CERTS makes n8n trust the mock's local certificate authority. The certificate and its key are created on your machine and never leave it.

Record, then replay

integration-mock start                # mode off: everything passes through
integration-mock packs enable slack   # only enabled services are ever intercepted
integration-mock record start         # run the workflow: real calls happen and are recorded
integration-mock record stop
integration-mock on                   # replay: the same calls are now answered from the recording
integration-mock log                  # what the workflow sent
integration-mock off                  # back to the real services

In replay, a call that nothing matches gets a 501. The real service is never contacted.

n8n Cloud

n8n Cloud does not let you set HTTP_PROXY or NODE_EXTRA_CA_CERTS, so proxy mode is not available.

Node	On Cloud
HTTP Request	Change the URL, or set the node's own Proxy option
A node whose credential has a base URL	Change the credential
A node with a fixed URL	Cannot be mocked

To use the HTTP Request node's own Proxy option, the node also needs Ignore SSL Issues. Set both, and clear both, with one command:

integration-mock creds swap workflow.json --via proxy --base-url https://your-mock --in-place
integration-mock creds swap workflow.json --via proxy --real --in-place

Important

Always clear it afterwards. A node left ignoring certificate errors while it calls a real service is less secure than before.