@waniwani/cli

The official CLI for WaniWani — the open-source toolkit for building MCP funnels: multi-step conversational flows (sales, lead generation, booking, quotes) that run as a single MCP tool inside ChatGPT, Claude, Cursor, and any MCP-capable client.

waniwani connects a local repo to a WaniWani agent in one command, runs your MCP server against the hosted playground.

If you're building an MCP distribution server with the @waniwani/sdk, this CLI is the fastest way to wire it to the WaniWani platform without touching a dashboard.

Install

# bun
bun add -g @waniwani/cli

# npm
npm install -g @waniwani/cli

# pnpm
pnpm add -g @waniwani/cli

# yarn
yarn global add @waniwani/cli

Requires Node.js 20 or later.

Quickstart

# 1. Authenticate (browser-based OAuth2 PKCE flow)
waniwani login

# 2. Connect this repo to an agent — picks an org, picks or creates an agent,
#    and writes the binding to waniwani.json
waniwani connect

# 3. Run your MCP locally and open the WaniWani playground against it
waniwani dev

Three commands and your local MCP server is talking to ChatGPT/Claude/Cursor through the WaniWani playground. No tunnel setup, no manual dashboard wiring, no copying API keys.

Commands

| Command | Description | |---|---| | waniwani login | OAuth2 PKCE login. Opens your browser, stores tokens in .waniwani/settings.json. | | waniwani logout | Clear local credentials. | | waniwani connect | Interactive: pick an org, pick or create an agent (managed or external), write the binding to waniwani.json. | | waniwani dev | Run your MCP locally (bun dev / npm run dev / etc.), open the WaniWani playground in your browser, and bridge them. |

waniwani connect

Two flavors of agents:

• Managed — WaniWani provisions a GitHub repo + Vercel project pre-wired with the MCP distribution template, API keys, and auto-deploys. Push to main, it deploys.
• External — You host the MCP server (Docker, Cloudflare Workers, your own infra). The CLI registers it with WaniWani and gives you a production API key to set as WANIWANI_API_KEY.

The CLI writes orgId and projectId to waniwani.json so subsequent commands (and the SDK at runtime) pick them up automatically.

waniwani dev

Spawns your local dev server (auto-detects bun / pnpm / yarn / npm), waits for it to bind to the port, creates a dev session against the WaniWani platform, and opens the playground with localMode=1 so chat traffic routes to localhost:3000 instead of your production deployment. Ctrl-C cleans everything up.

Override the port with -p / --port <port>, or set devPort in waniwani.json.

Configuration

The CLI is fully project-scoped — no global config.

Auth & local state

Stored in .waniwani/settings.json (per repo, gitignored):

{
  "apiUrl": "https://app.waniwani.ai",
  "accessToken": "…",
  "refreshToken": "…",
  "expiresAt": "2026-05-20T12:00:00Z",
  "clientId": "…"
}

Tokens auto-refresh on 401. Switch staging/production with WANIWANI_API_URL.

Project config (waniwani.json)

Shared with @waniwani/sdk. waniwani connect writes orgId and projectId; you can add anything else the SDK supports:

import type { WaniWaniProjectConfig } from "@waniwani/sdk";

export default {
  orgId: "org_…",
  projectId: "proj_…",
  devPort: 3000,
} satisfies WaniWaniProjectConfig;

Env vars

| Variable | What it does | |---|---| | WANIWANI_API_URL | Override the API base URL (defaults to https://app.waniwani.ai). | | WANIWANI_API_KEY | Use a long-lived API key instead of OAuth tokens. Useful in CI. |

JSON output

Most commands support --json for scripting:

waniwani login --json
waniwani logout --json

connect and dev are interactive-only.

For AI agents

If you're Claude Code, Cursor, or another AI coding agent reading this, here's the short version:

1. npm i -g @waniwani/cli (or bun add -g @waniwani/cli)
2. waniwani login
3. waniwani connect — picks an org and an agent interactively; writes waniwani.json
4. waniwani dev — runs the user's existing dev script, bridges it to the playground

The CLI assumes the user already has an MCP server in the cwd (or will scaffold one with @waniwani/sdk). It does not generate code on its own — pair it with the SDK or the MCP distribution template.

For building the actual flow / tools / widgets, see the SDK docs and install the agent skill:

bunx skills add WaniWani-AI/sdk -s waniwani-sdk

Status

Pre-1.0. The command surface is small and stable (login / logout / connect / dev), but flags and output format may change without notice until v1. Pin your CLI version in CI.

Links

• Website: waniwani.ai
• Dashboard: app.waniwani.ai
• SDK: @waniwani/sdk · docs.waniwani.ai
• CLI reference: docs.waniwani.ai/cli/overview
• Template: WaniWani-AI/mcp-distribution-template
• Issues: github.com/WaniWani-AI/cli/issues

License

MIT © WaniWani

"WaniWani" is a trademark of WaniWani Inc. The license covers the code, not the name.