Documentation Index
Fetch the complete documentation index at: https://docs.nudgen.net/llms.txt
Use this file to discover all available pages before exploring further.
Overview
Nudgen MCP gives AI agents tool-based access to your Nudgen workspace. Use it when your agent supports Model Context Protocol and you want it to call Nudgen tools directly instead of writing REST requests.Hosted server
Connect to Nudgen over HTTPS. No local MCP server process is required.
PAT auth
Use the same Personal Access Token as the Developer API.
Campaign tools
Create contacts, draft emails, create campaigns, and schedule launches.
Why this matters
MCP lets an AI assistant operate through structured Nudgen tools. That makes workflows like “create this contact, draft a campaign, schedule it for tomorrow, and report stats” easier for the agent to perform reliably. Use MCP when your agent supports Model Context Protocol tools. The Nudgen MCP server is hosted, so you do not need to run a local server process.Before you connect
- Create a Personal Access Token from Settings -> API Keys.
- Store the token as an environment variable, for example
NUDGEN_PAT. - Confirm your active workspace in Nudgen. MCP tools operate on the token owner’s active workspace.
Claude Code
Add Nudgen as an HTTP MCP server:.mcp.json:
Cursor
Open Cursor Settings -> Tools & MCP -> New MCP Server and add:Windsurf
Open the Windsurf MCP configuration and add the hosted HTTP server:Codex CLI
Add the server to your Codex MCP configuration:Generic MCP clients
Any MCP client that supports hosted HTTP MCP servers and custom headers can connect with this configuration:https://nudgen.net/api/mcp with the Authorization header.
Verify with JSON-RPC
You can verify the server without an MCP client:Available MCP tools
Available MCP tools mirror the core REST workflows:| Tool | REST equivalent |
|---|---|
get_current_user | GET /api/v1/user/me |
list_contacts | GET /api/v1/contacts |
create_contact | POST /api/v1/contacts/add |
list_campaigns | GET /api/v1/campaigns |
create_campaign | POST /api/v1/campaigns |
launch_campaign | POST /api/v1/campaigns/:id/launch |
get_campaign_stats | GET /api/v1/campaigns/:id/stats |
get_brand_settings | GET /api/v1/settings/brand |
update_brand_settings | PATCH /api/v1/settings/brand |
generate_email_draft | POST /api/v1/ai/generate |
Troubleshooting
| Symptom | What to check |
|---|---|
Unauthorized | Confirm the PAT is copied correctly, not revoked, and sent as Authorization: Bearer <token>. |
| Tools do not appear | Restart the MCP client and confirm it supports hosted HTTP MCP servers. |
Writes fail with 403 | Confirm the workspace plan and role can perform write actions. |
| Campaign does not send | Confirm the campaign worker is running and the campaign has eligible contacts. |
| Dashboard stats stay at zero | Test sends are excluded from launched campaign stats. Launch a real campaign or schedule one through launch_campaign. |
Security recommendations
- Create a separate PAT for each agent or environment.
- Revoke tokens when an agent, laptop, CI job, or workspace is no longer trusted.
- Keep manual approval enabled for write tools such as
create_campaign,launch_campaign, andupdate_brand_settings. - Use future
scheduledAtvalues when testing campaign launch workflows. - Avoid pasting raw PATs into chat logs, issue comments, or source files.