This page shows how to find influencers on X who post about your topic, so you can follow them, learn from them or reach out for a collaboration. One saved prompt in Claude or Cursor runs a Desearch search, groups the public posts by author and ranks the people who keep posting about your topic, with their follower counts. It is a first pass you can do without code in a few minutes.
The problem
Finding the right people to work with on X usually means scrolling search results, opening profiles one by one and keeping notes in a spreadsheet. It is slow, and it is easy to confuse someone who posted about your topic once with someone who keeps coming back to it. You want a short list of names, ranked, with their reach next to them.
What the app does
- Takes your topic words, such as
("sourdough" OR "bread baking"). - Searches public X posts on that topic with the Desearch
x-searchtool (up to 50 posts per search). - Groups the posts by author and keeps authors with 2 or more posts.
- Ranks them by total likes and shows followers, number of posts, total likes and their most liked post.
- With the saved prompt, also lists authors with 1 post and more than 10,000 followers as "worth a look".
- The script can hide handles and links when you share the result.
How to find influencers on X with this prompt
- Connect once. Add the hosted Desearch MCP server (
https://mcp.desearch.ai/mcp) to Claude or Cursor, with your Desearch API key in theAuthorizationheader. - Search. The saved prompt calls the
x-searchtool with your topic,lang:en -is:retweet, count 50, and theverifiedandmin_likessettings. - Group and rank. Your assistant groups the posts by author, keeps the people with 2 or more posts and ranks them by total likes, with follower counts from the posts.
- Check before you reach out. Open each profile. The table is a starting point, not a verdict.
The code
No code is needed for the saved prompt. For people who want a script, this is the core MCP call from find_influencers.py in the example (lines 88 to 98). The arguments come from build_args() in the same file, which adds -is:retweet and sets the count and the optional verified and min_likes filters.
async def search(args, key):
from mcp import ClientSession
from mcp.client.streamable_http import streamable_http_client
from mcp.shared._httpx_utils import create_mcp_http_client
async with create_mcp_http_client(headers={"Authorization": key}) as hc, \
streamable_http_client(MCP_URL, http_client=hc) as streams:
async with ClientSession(streams[0], streams[1]) as s:
await s.initialize()
res = await s.call_tool("x-search", args, read_timeout_seconds=180)
return parse_posts([c.model_dump() for c in res.content])The script reads your key from a file (~/.desearch-api-key, or --key-file PATH). If you keep your key in an environment variable such as DESEARCH_API_KEY, you can pass its value as key to search() in your own code.
The folder has saved-prompt.md (the no-code version), find_influencers.py (the script), README.md with a sample output, and offline tests in tests/.
Endpoints
MCP tool x-search
Example output
Saved sample from a search on 2026-10-16 for ("sourdough" OR "bread baking") lang:en with --verified --min-likes 10, handles hidden. This page shows that stored request and response. It does not run a live demo.
Sample request
build_args() sends this to the x-search tool:
{
"query": "(\"sourdough\" OR \"bread baking\") lang:en -is:retweet",
"count": 50,
"verified": true,
"min_likes": 10
}Sample response
| What came back | Result |
|---|---|
| Public posts | 50 |
| Authors | 38 |
| Authors with 2 or more posts | 5 |
The top three rows of the ranked table:
| # | Author | Followers | Posts | Likes |
|---|---|---|---|---|
| 1 | a creator #1 | 35,855 | 5 | 2,382 |
| 2 | a creator #2 | 2,036,399 | 3 | 1,877 |
| 3 | a creator #3 | 4,960 | 3 | 437 |
Two numbers matter most when you read it. Posts on the topic tell you who keeps coming back to it. Followers tell you the reach, and a creator with a few thousand engaged followers can be a better partner than a huge account.
Run it yourself
No code: connect the Desearch MCP server in Claude or Cursor, paste saved-prompt.md, and change the topic words.
Script:
git clone https://github.com/Desearch-ai/desearch-py-examples && cd desearch-py-examples/find-x-influencers-mcp
pip install -r requirements.txt
echo "<your key>" > ~/.desearch-api-key && chmod 600 ~/.desearch-api-key # key from console.desearch.ai
python find_influencers.py '("sourdough" OR "bread baking") lang:en' --verified --min-likes 10Useful flags: --verified and --min-likes (narrow the search), --min-posts (default 2), --hide-handles (share the table without handles or links). Keep OR terms inside parentheses.
What it costs
A 50-post search cost $0.0075, which is $0.00015 per post (the cost header returned by the same search through the Desearch API, GET /twitter with count 50, 2026-10-16 15:14 Asia/Tbilisi). See pricing.
Ideas to extend it
- Add a start date and end date in the saved prompt to see who posts about your topic now.
- Run it for two or three related topics and look for names that show up in more than one list.
- Export the ranked table to a sheet and add a column for notes after you open each profile.