n8n + SalesCrew
No dedicated node. The Webhook node catches SalesCrew events, and the HTTP Request node writes back through the API or an MCP tool with a scoped token.
- Webhook node in, HTTP node out
- No dedicated n8n node
- Self-hosted friendly

The short answer
- There is no SalesCrew node in the n8n catalogue. The connection is n8n's generic Webhook node for triggers and its HTTP Request node for writes, which is enough to build anything the app itself can do.
- Self-hosted n8n pairs well with SalesCrew's shape. Both run on infrastructure you control, so your workflow engine and your CRM database can sit in the same place rather than round-tripping through a vendor.
- Because SalesCrew's MCP server speaks MCP over HTTP, an n8n workflow with an AI step can call CRM tools the same way Claude or Cursor would, under a token you scope and revoke.
Mechanism
Generic nodes, specific behaviour.
| What | Mechanism | Direction | Objects | Status |
|---|---|---|---|---|
| Dedicated n8n node | Does not exist | , | , | Not available |
| SalesCrew to n8n | Outbound webhook to an n8n Webhook node URL | Outbound | Any registered event | Available |
| Delivery reliability | Deliveries logged with retries and a rotatable secret | Outbound | Webhook deliveries | Available |
| n8n to SalesCrew | HTTP Request node against the SalesCrew API | Inbound | Contacts, companies, deals, tasks, notes | Available |
| n8n agent to SalesCrew | MCP over HTTP with a scoped token | Inbound | Every action the UI has | Available |
| Community node | None published by us | , | , | Not available |
Mechanisms read from the SalesCrew integration table, September 2026. n8n's own docs are the authority on node behaviour and on production webhook URLs versus test ones.
Setup
The one trap is using the test URL in production.
- 1
Add a Webhook node
Create the workflow, add the Webhook node, and take the production URL rather than the test URL. The test URL only listens while the editor is open.
- 2
Register it in SalesCrew
webhook_create with the URL and the events. webhook_test fires a sample payload so the workflow can be built against real JSON.
- 3
Activate the workflow
An inactive workflow does not answer on the production URL, which is the most common reason a first delivery fails.
- 4
Add the write path
An HTTP Request node against the API, with a token from mcp_token_issue scoped by channel, profile and area.
- 5
Or point an AI step at MCP
The MCP server is served per instance over HTTP, so an n8n AI agent step can use the same tools a person's screen uses.
- 6
Check deliveries after go-live
webhook_deliveries lists what was sent and what failed; webhook_retry replays a single delivery.
The accurate word is webhook
Questions
- Will you publish an n8n node?
- It is not on the published roadmap. Webhooks, the API and the MCP server are the supported connection.
- Does self-hosting n8n change anything?
- Only for the better. SalesCrew already runs one Supabase project per client, so keeping the workflow engine on your own infrastructure keeps the whole path under your control.
- Can an n8n AI agent operate the CRM?
- Yes, over MCP with a scoped token. It gets the same permissions as a person with that scope, external sends still queue for approval, and every call is audited.
- My first webhook never arrived.
- Check three things in order: the workflow is active, you used the production URL, and webhook_deliveries shows the attempt. That covers nearly every case.
- Is there a rate limit?
- Not published. Build with retries and idempotent writes, which is the right shape regardless of any vendor's limit.
Point one workflow at your own instance
Webhook node in, HTTP node out, and an MCP token that you can revoke in a click.