Find X Posts for a Prompt with Python

X Post Links Search is a short Python example. It takes a prompt and a count, asks for X posts that match, and prints them. Each post can include text, a URL, engagement counts and a short summary description.

The problem

A keyword search on X returns a timeline. When you want posts for a prompt, with the URL and the counts in one object, a single links call is the smaller tool.

What the example does

• Sends a prompt and a count.
• Prints the posts that came back.

How it works

The file calls twitter_links_search with prompt and count. The current SDK method with those same arguments is ai_x_links_search. It posts to POST /desearch/ai/search/links/twitter.

This is separate from GET /twitter, which takes an X query string, sort and date filters. Use that endpoint when you need operators such as since: or min_faves:.

The code

result = desearch.twitter_links_search(prompt="Bittensor", count=10)

Set the key in the environment when you use the current client. The file's setup lines use a placeholder string. Do not commit a real key.

Endpoints

• POST /desearch/ai/search/links/twitter

Example output

The file documents post fields: text, URL, reply count, retweet count, like count, view count, quote count and a summary description. It does not include a saved sample run or a measured cost.

Run it yourself

The file linked above is twitter_links_search.py in the examples repository. Install desearch-py, create a key at console.desearch.ai, and call ai_x_links_search with a prompt and count.

What it costs

This example does not record X-Desearch-Cost-Usd. Current responses include that header. See pricing.

Ideas to extend it

• Write the posts to the CSV exporter's columns.
• Follow the loudest post into the replies example.
• Combine the posts with web links for the same prompt.

{
  "mcpServers": {
    "desearch": {
      "url": "https://mcp.desearch.ai/mcp",
      "headers": {
        "Authorization": "Bearer <DESEARCH_API_KEY>"
      }
    }
  }
}

Use Desearch to find X posts for the prompt I give you. Return the posts you received with text, URL and engagement counts. Do not add posts you did not retrieve.