docs(README.md): adding mcp-meme-sticky to README

[nkapila6]

Description

A MCP server that allows you to generate memes and further convert them to stickers using the open source free memegen API. No other API use!

Server Details

• Server: New MCP
• Changes to: N/A

Motivation and Context

While there exist a few meme generator MCPs, this one adds extra features, namely:

• Allows users to create Telegram stickers from generated meme. (WhatsApp coming soon).
• Telegram stickers are generated using my custom MCP-Sticky-Tele Telegram bot.
• Gives direct meme link to user and can save generated meme to desktop. UPDATE: It opens the link now directly!
• GIF support and WhatsApp coming in a few days!
• Not limited to memegen templates and can generate memes using images on the internet!

How Has This Been Tested?

Have tested this on Claude using both 3.7 and 3.5 models. Works fine with both, below are some screenshots.

Telegram sticker generation using my custom built Telegram bot - https://github.com/nkapila6/mcp-sticky-tele

Meme saved on desktop!

Breaking Changes

Yes, the user will need to update their MCP configuration on their MCP host of choice, eg. Claude, Cursor, Goose, etc.

UPDATE: added installation instructions in README.

Types of changes

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to change)
• Documentation update

Checklist

• I have read the MCP Protocol Documentation
• My changes follows MCP security best practices
• I have updated the server's README accordingly
• I have tested this with an LLM client
• My code follows the repository's style guidelines
• New and existing tests pass locally
• I have added appropriate error handling
• I have documented all environment variables and configuration options

Additional context

One important design decision I had to make as Claude Desktop does not support MCP Sampling yet.

To resolve this, I had to a few tool calls (instructing LLM via documentation). I will experiment more on sampling differently until Claude supports it or I find a better solution.

UPDATE: I preprocessed the templates and created metadata for it. Using cosine similarity (dot products) to fetch the best template since nested tool calling is not reliable. Works well, much more deterministic and stable!

[olaservo]

Thanks for your contribution to the servers list. This has been merged in this combined PR: #2007

This is a new process we're trying out, so if you see any issues feel free to re-open the PR and tag me.