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.

Connecting Workflows

Four ways to connect S6S workflows to your website, app, or service.

Webhooks

Your website's backend POSTs data to a unique URL. Best for forms, notifications, and event-driven flows.

REST API

Programmatically deploy and trigger workflows from any backend. Best for server-to-server integration.

Scheduled (Cron)

Run workflows on a recurring schedule. Best for reports, monitors, and data syncs.

SDK & MCP

TypeScript SDK for apps, MCP server for AI agents (Claude, Cursor). Best for developers.

Webhooks

The simplest way to connect a workflow to a website. Deploy a workflow with a webhook trigger, get a unique URL, and POST data to it from your backend.

1. Deploy with a webhook trigger

cURL

curl -X POST https://s6s.ai/api/v1/workflows/deploy \
  -H "Authorization: Bearer s6s_YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "Contact Form Handler",
    "trigger": {
      "type": "webhook",
      "path": "/contact-form",
      "method": "POST"
    },
    "steps": [
      {
        "ref": "classify",
        "action": "ai",
        "config": {
          "systemPrompt": "Classify the message as sales, support, or spam. Reply with one word.",
          "prompt": "From: {{trigger.output.body.name}}\nMessage: {{trigger.output.body.message}}"
        }
      },
      {
        "ref": "notify",
        "action": "slack",
        "config": {
          "channel": "#leads",
          "textTemplate": "New {{classify.output.text}} message: {{trigger.output.body.message}}"
        },
        "credential": "default"
      }
    ],
    "flow": ["classify", "notify"]
  }'

2. Get your webhook URL

The deploy response includes your unique URL and a secret for authentication (abridged):

{
  "ok": true,
  "workflowId": "8c1f2a4e-5b6d-4e7f-9a0b-1c2d3e4f5a6b",
  "runId": null,
  "webhookTrigger": {
    "path": "/contact-form",
    "invokeUrl": "https://s6s.ai/api/webhooks/contact-form",
    "secret": "3f9a1c7e5b2d8f4a6c0e9b3d7f1a5c8e2b6d0f4a9c3e7b1d"
  }
}

The secret is 48 hex characters. It is returned only on the first deploy; later deploys return "secret_hint": "previously_generated" instead, so store it. Deploying a webhook workflow does not start a run. If a credential is missing (here, Slack), the response lists it under missing_credentials and has no webhook details yet; connect it and deploy again.

3. Call it from your website's backend

The secret lets anyone start runs, so send it from your server, not from front-end code that visitors can read.

JavaScript (server-side)

const response = await fetch(
  "https://s6s.ai/api/webhooks/contact-form",
  {
    method: "POST",
    headers: {
      "Content-Type": "application/json",
      "X-Webhook-Token": process.env.S6S_WEBHOOK_SECRET
    },
    body: JSON.stringify({
      name: form.name,
      email: form.email,
      message: form.message
    })
  }
);

Webhook authentication

Five methods are supported (use any one). All compare against the secret in constant time.

Method	Where	Description
Direct token	X-Webhook-Token	The secret as the header value
Query parameter	?token=	The secret in the URL, for callers that cannot set headers. URLs can end up in logs, so prefer a header
HMAC signature	X-Webhook-Signature	HMAC-SHA256 of the raw request body, keyed with the secret (hex)
Bearer	Authorization: Bearer	The secret as a Bearer token
Telegram	X-Telegram-Bot-Api-Secret-Token	The secret set as Telegram's webhook secret_token

Data available to steps

// Any step after a webhook trigger can read:
{{trigger.output.body}}          // Parsed JSON body
{{trigger.output.body.name}}     // Specific field
{{trigger.output.headers}}       // Request headers, minus Authorization, Proxy-Authorization, Cookie, X-Webhook-Token, X-Webhook-Signature and X-Telegram-Bot-Api-Secret-Token
{{trigger.output.query}}         // Query string parameters (token removed)
{{trigger.output.method}}        // HTTP method
{{trigger.output.path}}          // Webhook path
{{trigger.output.receivedAt}}    // ISO timestamp

REST API

For backends and scripts. Deploy a workflow, trigger it programmatically, and get results back via polling, streaming, or callbacks.

Run an existing workflow

curl -X POST https://s6s.ai/api/v1/workflows/{workflowId}/run \
  -H "Authorization: Bearer s6s_YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{ "input": { "company": "Acme Corp" } }'

The response returns the new runId with status running. The input becomes {{trigger.output.body}} for webhook workflows and {{trigger.output.payload}} for manual ones. A raw JSON body without the input wrapper also works.

Getting results back

Method	How it works
Polling	`GET /api/v1/runs/{runId}` — status, timing and step outputs
Streaming	`GET /api/runs/{runId}/stream` — Server-Sent Events with real-time step progress (Bearer API key accepted)
Callback	Pass `callback_url` on deploy — if the deploy starts a run, S6S POSTs `{ run_id, status, timestamp }` when it ends (up to 3 attempts). Outputs are not included; poll for them
Wait	Pass `waitForCompletion: true` on deploy — if the deploy starts a run, the response waits up to 30s and includes `run_result`

Deploy starts a run right away only for a manual trigger (pass autoRun: true to run a webhook or telegram workflow on deploy; a schedule never runs on deploy), and not while a credential is missing. See API Keys for authentication and rate limits.

Scheduled (Cron)

For recurring jobs like daily reports, periodic monitors, or data syncs. No external trigger needed — S6S runs the workflow on your schedule.

Recipe JSON

{
  "name": "Daily Competitor Report",
  "trigger": {
    "type": "schedule",
    "cron": "0 9 * * 1-5",
    "timezone": "America/New_York"
  },
  "steps": [
    {
      "ref": "research",
      "action": "ai",
      "config": {
        "agentMode": true,
        "systemPrompt": "You research competitor pricing pages and summarize what they offer.",
        "prompt": "Read https://example.com/pricing and summarize the current plans and prices."
      }
    },
    {
      "ref": "send_report",
      "action": "email",
      "config": {
        "to": "[email protected]",
        "subject": "Daily Competitor Report",
        "bodyTemplate": "{{research.output.text}}"
      },
      "credential": "default"
    }
  ],
  "flow": ["research", "send_report"]
}

With agentMode on, the AI step gets an http_request tool automatically, so it can read web pages. Its answer is in {{research.output.text}}. The email step uses your connected email (SMTP) credential.

Deploying a schedule workflow does not start a run. S6S checks schedules every 60 seconds and fires each schedule at most once per minute. It skips a run when the workflow still has an active run or the account has insufficient credits.

See Triggers for cron expression syntax and timezone details.

SDK & MCP Server

TypeScript SDK

TypeScript

import { S6SClient } from "@s6s/sdk";

const client = new S6SClient({
  baseUrl: "https://s6s.ai",
  apiKey: process.env.S6S_API_KEY!,
});

const recipe = {
  name: "Hello",
  trigger: { type: "manual" },
  steps: [{ ref: "greet", action: "transform", config: { template: { message: "Hello" } } }],
  flow: ["greet"],
};

// Deploy: a manual-trigger workflow starts a run right away
const result = await client.deployWorkflow(recipe, { waitForCompletion: true });
console.log(result.runId, result.run_result);

// Run it again with input, then stream progress
const { runId } = await client.runWorkflow(result.workflowId, { name: "Ada" });
for await (const event of client.streamRun(runId)) {
  console.log(event.event, event.data);
}

// Manage runs
const status = await client.getRunStatus(runId);
await client.cancelRun(runId);   // active runs only
await client.retryFailed(runId); // failed runs only

The SDK is not on npm yet; the SDK page shows how to use it from the repository. Full SDK docs: TypeScript SDK

MCP Server (for AI agents)

The MCP server exposes 15 tools so Claude, Cursor, or any MCP-compatible agent can deploy and manage workflows directly.

npx @s6s/mcp-server

Full MCP docs: MCP Server

Practical Examples

Webhook

Contact form + AI + Slack

Your website's backend POSTs the form to a webhook. An AI step classifies it (sales, support or spam) and posts it to Slack.

Webhook

GitHub PR + Notify

GitHub sends a webhook when a PR opens (put the secret in the URL as ?token=). The workflow checks the PR details and posts a summary to Slack.

Schedule

Daily competitor monitor

Runs at 9 AM on weekdays. An AI agent reads competitor pricing pages, writes a digest, and emails it to the team.

REST API

Lead enrichment API

Your CRM calls the run endpoint with the lead's email. The workflow enriches and scores the lead; your CRM polls or streams the run for the result.

Ready to connect?

Deploy your first workflow and get a webhook URL in seconds.

Quick Start API Keys