Appie.run

Appie.run MCP server

Find Reddit threads and draft replies, track LinkedIn profiles and leads, write on-brand articles, run AI chatbots and create trackable short links from Cursor, Claude or any other Model Context Protocol client.

Appie.run is an AI marketing stack that learns your brand from your website. This hosted MCP server gives your AI client the same brand context, so the content and answers it produces sound like you.

Try it free

New Appie.run accounts get 500 free credits, no credit card needed. That's enough for about 5 light articles or 500 chatbot replies. Read tools never spend credits.

Quick start

1. Sign up at Appie.run and open your project.
2. Go to API in the project sidebar and generate a key (it starts with esk_). Only project admins can do this.
3. Add the server to your MCP client config:

{
  "mcpServers": {
    "appie": {
      "type": "streamable-http",
      "url": "https://appie.run/mcp",
      "headers": { "Authorization": "Bearer YOUR_APPIE_API_KEY" }
    }
  }
}

The same config is in examples/mcp-config.json. Never commit a real key.

Grok Build plugin

This repo is also a Grok Build plugin. .mcp.json connects to https://appie.run/mcp and sends your key from the APPIE_API_KEY environment variable as a Bearer token. Set it before starting Grok:

export APPIE_API_KEY=esk_your_key

The plugin ships no code or hooks. It includes the five instruction-only agent skills in skills/ (see Agent Skills), which call the same MCP server. Its only network endpoint is https://appie.run/mcp, and its only credential is APPIE_API_KEY.

What you can do

• Find Reddit conversations worth joining. Save keywords and competitors, list matching Reddit threads, and draft a reply for any thread. Appie never posts to Reddit; you paste the reply yourself.
• Track LinkedIn profiles and leads. Add the LinkedIn profiles you want to track and list the leads, with email and phone when a lookup has found them.
• Write articles in your voice. Start a light (~500 words) or deeply researched (~3,000 words) article using an author profile trained on your writing samples, then check its status until the draft is ready.
• Manage brand context. Read and update the project Brain: your website, about text, brand voice, key messages, ideal customer and competitors.
• Build and query chatbots. Create a chatbot, train it on text, URLs, sitemaps or your Appie articles, and ask it questions.
• Create and track short links. Make links on s.appie.run and see clicks and unique visitors for the last 30 days.

Try asking your AI client

• "Write a light article on onboarding emails for SaaS in our brand voice, and tell me when it's done."
• "Create a support chatbot and train it on our sitemap."
• "Make a short link for our pricing page and show me last month's clicks."
• "Find Reddit threads where people ask for alternatives to our competitors, and draft a reply for the best one."
• "Track this LinkedIn profile and list the latest leads."

Tools

21 tools across short links, articles, the Brain, chatbots, Reddit and LinkedIn.

Tool	What it does	Cost
create_short_link	Creates a short link on s.appie.run, with an optional custom code	1 credit
link_stats	Clicks, uniques and a 30-day series for one link	Free
create_article	Starts a light or deeply researched article in an author's voice (async)	100 or 500 credits
article_status	Status and content of one article	Free
list_authors	Lists author profiles and their samples	Free
create_author	Creates an author, optionally with a first writing sample	See docs
add_author_sample	Adds a text or blog-URL sample to an author (up to 5)	See docs
get_brain	Reads the project's brand context	Free
update_brain	Updates selected brand context fields	See docs
list_chatbots	Lists chatbots and their training status	Free
create_chatbot	Creates a chatbot with optional prompt, model and temperature	Free
add_chatbot_content	Adds text, URL, sitemap or article knowledge to a chatbot	See docs
chatbot_status	Training status and recent sources for one chatbot	Free
ask_chatbot	Sends a message to a chatbot and returns its reply	1 credit
get_reddit_settings	Reads Reddit keywords, competitors and reply guidance (seeded from the Brain if nothing is saved)	Free
update_reddit_settings	Sets Reddit keywords, competitors or reply guidance (up to 25 items per list)	Free
list_reddit_mentions	Lists Reddit threads for saved keywords and competitors, with the latest reply draft	Free to list; a scan runs at most once a day for 10 credits
draft_reddit_reply	Drafts a reply for one Reddit thread; you post it yourself	2 credits
get_linkedin_profiles	Lists the LinkedIn profiles the project tracks	Free
set_linkedin_profile	Adds or updates a tracked LinkedIn profile	Free to save; scheduled scans cost 30 credits
list_linkedin_leads	Lists leads, with email and phone once a lookup has found them	Free to list; lookups cost 5 credits per lead

Every tool result includes credits_remaining. Writes spend the project's credits first, then the project owner's.

Permissions and errors

A key can only call tools for apps the person who generated it is allowed to use. Project owners can use every app. Regenerating a key revokes the old one immediately.

Error	Meaning
HTTP 401	Missing or revoked key
app_not_allowed	The key's owner can't use that app
insufficient_credits	Not enough credits (includes required and available)
author_profile_required / writing_sample_required	Add an author with a writing sample before create_article
not_found	The link or article isn't in this project

Agent Skills

This repo also ships five agent skills in skills/:

• appie-reddit-marketing: find Reddit threads for your keywords and competitors and draft replies you post yourself
• appie-linkedin-growth: track LinkedIn profiles and list leads
• appie-article-writer: light or deeply researched articles in your voice
• appie-chatbot: build, train and chat with project chatbots
• appie-url-shortener: short links and click stats

Install them with:

npx skills add appierun/appie-mcp

Or in Claude Code:

/plugin marketplace add appierun/appie-mcp
/plugin install appie-skills@appie

Documentation

Full reference: appie.run/docs/mcp

License

MIT, see LICENSE.