For AI Agents
FunnelFizz is built to be set up by an agent, not just with one. Three artifacts make that work:
| Tool | Purpose | Where | |
|---|---|---|---|
| 🚀 | Skill file | One-shot setup runbook the agent follows step-by-step | /skill.md |
| 🤖 | MCP server | Tools the agent calls to set up + read your funnel | https://mcp.funnelfizz.com |
| 📖 | llms.txt + Context7 | Machine-readable docs for any docs question | /llms.txt · /llms-full.txt |
When to use which
skill.md, start here. Paste its URL into your agent and say "follow this." It walks the agent through signup, tracking install, Stripe connect, providers, identify calls, and your first automation, calling MCP for everything that has an MCP tool, asking you only for the decisions that need you (your domain, framework, etc.). See Setup.
MCP, for ongoing work. After setup, your agent can do everything you can do in the dashboard — build funnels, send emails, activate automations, manage team and billing — via ~120 tools. Safety tiers (T0 reads / T1 mutations / T2 admin) protect destructive actions with target_tokens and code-paste confirmation. See MCP and Safety tiers.
llms.txt + Context7, for docs questions. Sparse index, dense dump, or indexed search. The agent reads docs directly without leaving the conversation. See Reading docs.
The shortest possible setup
For agents that support MCP (Claude Desktop, Claude Code, Cursor, Windsurf, ChatGPT, Zed):
- Connect MCP. Config block pointing at
https://mcp.funnelfizz.comwith your API key. How. - Tell the agent: "Read https://docs.funnelfizz.com/skill.md and follow it."
- Answer the questions it asks.
Five minutes from "I have an account" to "tracking live, Stripe connected, providers wired."