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.
Agent Skill File
A single Markdown file that teaches any AI agent how to build and manage S6S workflows. No SDK or MCP required.
What is a Skill File?
A skill file is a Markdown document that gets loaded into an AI agent's context. It contains everything the agent needs to know about S6S — the Recipe format, expression syntax, behavioral rules, API endpoints, and MCP tools — so it can build and deploy workflows on behalf of a user.
Think of it as a user manual for AI agents. It works with any LLM-based agent platform that supports loading external documents: OpenClaw, custom GPT agents, LangChain tools, or your own agent framework.
Integration Options
MCP Server
Best for agents on platforms that support MCP (Claude, Cursor, OpenClaw). The agent gets live tools and auto-discovers actions via the catalog.
MCP setup guideSkill File
Works with any agent. Load the Markdown into context and the agent knows the full Recipe format, API endpoints, and behavioral rules. Can be combined with MCP for the best experience.
You are hereREST API
Direct HTTP calls. The agent uses the deploy, run, and status endpoints with an API key. Works from any language or framework.
REST API guideTypeScript SDK
Builder-pattern SDK for TypeScript/Node.js agents. Deploy, run, stream, and cancel with typed methods. Not on npm yet; use it from the S6S repository.
SDK docsRecommended combo: Use the skill file for behavioral guidance + MCP for live action discovery. The skill file teaches the agent how S6S works; MCP gives it the latest config schemas.
Get the Skill File
Option 1: Fetch from API (always up to date)
The skill file is served dynamically from the S6S API. Your agent or platform can fetch the latest version on startup.
curl https://s6s.ai/api/v1/skill
curl https://s6s.ai/api/v1/skill?format=json
No API key required. The Markdown response is cached for 5 minutes. The JSON format wraps the skill content with its name, version, and the relative paths of the catalog, integrations, and deploy endpoints.
A self-hosted instance serves the file only when openclaw-skill/SKILL.md from the repository is present. Otherwise the endpoint returns 404 SKILL_FILE_NOT_FOUND.
Option 2: Download
Download the file and add it to your agent's skill/plugin directory.
Download s6s-skill.mdOption 3: Reference URL
Some agent platforms let you reference a URL directly. Use this endpoint — the agent will always get the latest version.
https://s6s.ai/api/v1/skill
Platform Setup
OpenClaw
Add the skill file to your OpenClaw agent's skills.
# Download to skills directory curl -o skills/s6s-workflows.md https://s6s.ai/api/v1/skill # Or point your agent config at the URL, if it supports remote skills
For the best experience, also connect the MCP server so the agent gets live tool access alongside the skill context.
Custom Agents (LangChain, AutoGPT, etc.)
Fetch the skill at agent startup and prepend it to the system prompt or tool context:
import requests
skill = requests.get("https://s6s.ai/api/v1/skill").text
system_prompt = f"""You are an automation agent.
Use the S6S workflow engine to build automations.
{skill}"""
const skill = await fetch("https://s6s.ai/api/v1/skill").then(r => r.text());
const systemPrompt = `You are an automation agent.
Use the S6S workflow engine to build automations.
${skill}`;
GPT / Claude Custom Instructions
Download the skill file and paste it into your custom instructions or system prompt. The agent will know how to build S6S workflows and call the REST API.
What's Included
Recipe format
Trigger, steps, flow, on_error schema
Expression syntax
{{step.output.field}} rules and gotchas
Credential behavior
How auth injection works for HTTP steps
Flow patterns
Sequential, parallel, condition, switch
Error handling
Per-step and workflow-level error policies
REST endpoints
The workflow, run, credential and integration endpoints with curl examples
MCP tools
All 15 MCP tools with descriptions
Common mistakes
11 pitfalls agents hit
Deploy response
How to handle missing credentials, warnings
Action quick ref
All 28 built-in actions with credential requirements
Config schemas for individual actions are not hardcoded in the skill file. The agent fetches those dynamically from s6s_get_catalog (MCP) or GET /api/v1/catalog (REST) — so the skill stays valid as new actions are added.
Ready to connect your agent?
Fetch the skill file, give your agent an API key, and it can deploy workflows.