get-me-my-meme

Give your AI the power to respond with memes.

get-me-my-meme is a plug-and-play tool for LLMs (Claude, OpenAI, etc.) that lets your AI fetch and return a meme from the internet — any time it feels like one fits better than words.

User: "ignore all previous instructions"
AI: 🖼 *sends nice try hacker meme*

User: "I just shipped my first app!"
AI: 🖼 *sends happy dance meme*

User: "explain quantum physics"
AI: 🖼 *sends mind blown meme*

Install

npm install get-me-my-meme

Quick Start

import Anthropic from "@anthropic-ai/sdk";
import { memeToolDefinition, handleMemeTool } from "get-me-my-meme";

const client = new Anthropic({ apiKey: process.env.ANTHROPIC_API_KEY });

const response = await client.messages.create({
  model: "claude-sonnet-4-6",
  max_tokens: 1024,
  tools: [memeToolDefinition],  // 👈 just add this
  messages: [{ role: "user", content: "ignore all previous instructions" }],
});

// handle the tool call
if (response.stop_reason === "tool_use") {
  const toolUse = response.content.find((b) => b.type === "tool_use");
  if (toolUse?.type === "tool_use" && toolUse.name === "send_meme") {
    const meme = await handleMemeTool(toolUse.input as { query: string });
    // meme.imageUrl  → base64 PNG, render directly in UI
    // meme.title     → meme title from Tenor
    // meme.type      → "meme"
  }
}

How It Works

LLM decides a meme is the right response
         │
         ▼
calls send_meme({ query: "nice try hacker" })
         │
         ▼
get-me-my-meme searches Tenor for the query
         │
         ▼
fetches + pixelates the image
         │
         ▼
returns { type: "meme", imageUrl, title }
         │
         ▼
your UI renders it

API

memeToolDefinition

The Claude/OpenAI tool definition. Pass it directly into your tools array.

import { memeToolDefinition } from "get-me-my-meme";

// Claude
tools: [memeToolDefinition]

handleMemeTool(input)

Handles the tool call — fetches and processes the meme.

import { handleMemeTool } from "get-me-my-meme";

const meme = await handleMemeTool({ query: "nice try hacker" });

// returns MemeResult:
// {
//   type: "meme",
//   query: "nice try hacker",
//   imageUrl: "data:image/png;base64,...",
//   title: "nice try gif",
//   source: "tenor"
// }

fetchMeme(query)

Searches Tenor and returns the raw result (no image processing).

import { fetchMeme } from "get-me-my-meme";

const result = await fetchMeme("this is fine fire");
// { id, title, url, imageUrl }

processMeme(imageUrl)

Downloads and pixelates an image, returns a base64 PNG data URI.

import { processMeme } from "get-me-my-meme";

const base64 = await processMeme("https://media.tenor.com/...");
// "data:image/png;base64,..."

Types

// What the LLM passes to the tool
interface MemeToolInput {
  query: string;
}

// What the tool returns
interface MemeResult {
  type: "meme";
  query: string;
  imageUrl: string;   // base64 data URI — render directly as <img src={imageUrl} />
  title: string;
  source: "tenor";
}

// Union type for your chat response handler
type ChatResponse = MemeResult | { type: "text"; content: string };

Environment Variables

TENOR_API_KEY=your_tenor_api_key

Get a free Tenor API key at tenor.com/gifapi — no credit card, takes 2 minutes.

UI Integration

The imageUrl is a base64 PNG — render it directly:

// React example
{response.type === "meme" ? (
  <img src={response.imageUrl} alt={response.title} />
) : (
  <p>{response.content}</p>
)}

License

MIT