Developer section: these pages cover the S6S workflow engine for self-hosted installs and developers. On s6s.ai, API keys and MCP access are issued by S6S on request (contact us); on a self-hosted install, an admin creates them on the API Keys page. For the AI Visibility product, start at the AI Visibility docs.

MCP Server

Give any MCP-compatible AI agent the ability to deploy, monitor, and manage workflows.

What is MCP?

The Model Context Protocol (MCP) is an open standard that lets AI agents use external tools. The S6S MCP server gives your agent workflow superpowers — deploy persistent multi-step automations, check run status, list connected credentials, and more. All from natural language.

Works with Claude Desktop, Claude Code, Cursor, OpenClaw, and any MCP-compatible client.

Quick Start

1. Get an API key

On s6s.ai, API keys and MCP access are issued by S6S on request (contact us). On a self-hosted install, an admin creates a key on the API Keys page. See API Keys for details.

2. Install

npm install -g @s6s/mcp-server

The package installs an s6s-mcp command. Or run it directly without installing:

npx @s6s/mcp-server

3. Configure your AI agent

Add S6S to your agent's MCP configuration. See the platform-specific examples below.

Setup by Platform

Claude Desktop

Add to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):

claude_desktop_config.json

{
  "mcpServers": {
    "s6s": {
      "command": "npx",
      "args": ["@s6s/mcp-server"],
      "env": {
        "S6S_API_KEY": "s6s_YOUR_API_KEY",
        "S6S_BASE_URL": "https://s6s.ai"
      }
    }
  }
}

Claude Code

Add to .mcp.json in your project root:

.mcp.json

{
  "mcpServers": {
    "s6s": {
      "command": "npx",
      "args": ["@s6s/mcp-server"],
      "env": {
        "S6S_API_KEY": "s6s_YOUR_API_KEY",
        "S6S_BASE_URL": "https://s6s.ai"
      }
    }
  }
}

Cursor

Add to .cursor/mcp.json in your project:

.cursor/mcp.json

{
  "mcpServers": {
    "s6s": {
      "command": "npx",
      "args": ["@s6s/mcp-server"],
      "env": {
        "S6S_API_KEY": "s6s_YOUR_API_KEY",
        "S6S_BASE_URL": "https://s6s.ai"
      }
    }
  }
}

OpenClaw

Add the same server entry to your OpenClaw MCP configuration:

OpenClaw MCP config

{
  "mcpServers": {
    "s6s": {
      "command": "npx",
      "args": ["@s6s/mcp-server"],
      "env": {
        "S6S_API_KEY": "s6s_YOUR_API_KEY",
        "S6S_BASE_URL": "https://s6s.ai"
      }
    }
  }
}

Environment Variables

Variable	Required	Description
S6S_API_KEY	Yes	Your S6S API key (starts with `s6s_`). Without it the server starts with a warning, and every tool except the catalog and integration lookups fails with 401.
S6S_BASE_URL	No	API base URL. Defaults to `https://s6s.ai`. Set this if you are self-hosting.

Available Tools

The MCP server exposes 15 tools, 2 prompts and 1 resource. Each tool carries MCP annotations: read-only tools only read data, and destructive tools cancel or delete. Your MCP client decides how to use these hints, for example when to ask for confirmation.

Tool	Description	Annotations
s6s_get_catalog	Get the catalog: node types with config schemas, triggers, expression and flow syntax. Cached for 5 minutes	read-only
s6s_deploy_workflow	Deploy a Recipe, raw IR, or n8n export (as a JSON string). Starts a run right away only for a manual trigger, and not while a credential is missing. Optional waitForCompletion (up to 30 s) and callback_url	—
s6s_validate_workflow	Pre-flight check of a deployed workflow: credentials, config, schedule, structure. Returns ready, warnings or blocked	read-only
s6s_run_workflow	Start a run for a deployed workflow, with optional JSON input	—
s6s_get_run_status	Get run status, timing, errorCode, failureReason and per-step results	read-only
s6s_get_workflow	Get one workflow by ID, including its deployed definition (workflow IR)	read-only
s6s_list_workflows	List your workflows (up to 50)	read-only
s6s_list_runs	List recent runs, with optional status filter and limit (default 50, max 200)	read-only
s6s_cancel_run	Cancel an active run	destructive
s6s_retry_run	Retry a failed run (the same run restarts)	—
s6s_delete_workflow	Delete a workflow. Its schedule stops and its active runs are cancelled	destructive
s6s_update_workflow	Rename a workflow or enable/disable it	—
s6s_list_credentials	List your credential bindings and their status (no secrets exposed)	read-only
s6s_list_integrations	List built-in actions, trigger types and credential providers	read-only
s6s_get_integration	Get details for a built-in action, a credential provider, or a stored integration	read-only

Prompts

The server also exposes MCP prompts — guided workflows that help your agent accomplish common tasks.

Prompt	Description	Arguments
create-workflow	Guided workflow creation — step-by-step help from catalog to deploy	`description` — what the workflow should do
debug-run	Systematic debugging of a failed or stuck run	`run_id` — the run ID to debug

Resources

URI	Description
s6s://catalog	Full workflow catalog (actions, triggers, expressions, flow patterns). Shares a 5-minute cache with `s6s_get_catalog`.

Example

Tell your AI agent:

"Create a workflow that fetches the Hacker News front page every hour and posts the top story to Slack"

The agent will:

Call s6s_get_catalog to learn available node types
Build a Recipe with a schedule trigger, HTTP fetch, transform, and Slack post
Call s6s_deploy_workflow to deploy it. If Slack is not connected yet, the response lists it under missing_credentials with a connect_url
S6S runs it every hour on the schedule. The agent can check results with s6s_list_runs and s6s_get_run_status

Self-Hosting

If you are running S6S on your own infrastructure, set S6S_BASE_URL to your instance URL in the MCP configuration:

"env": {
  "S6S_API_KEY": "s6s_YOUR_API_KEY",
  "S6S_BASE_URL": "https://workflows.mycompany.com"
}

See Self-Hosting for full deployment instructions.

Ready to connect your agent?

Get an API key, then add the server to your MCP client.

Get an API Key