I tried to find MongoDB connection strings over 1000 public GitHub repositories

I tried to see if I could get other people Mongo Database connection string by just searching for it on GitHub search. Yes, I found a few.

I tried connecting to a few and yes, it worked!

Before you call the cops on me, listen to my backstory. 🤗

I was working on a NodeJS/Express application for practice and I remembered I pushed the .env file to my remote repository. While working on fixing this error, I thought about how many people would have made this error and it is going to stay somewhere in the commit histories even if the secrets eventually get unstaged. So I took the bait and made this GitHub search. While most of the results are not an actual connection string, a good number of them are still alive and functional.

[DISCLAIMER: NO HARM INTENDED, THIS IS JUST TO CREATE A PUBLIC AWARENESS]

How I scanned through the 1000 repositories

Actually, GitHub Search API limits to 1,000 results for each search. Using the scripts below, I was able to generate repositories whose code included mongodb+srv:

// index.ts
import dotenv from "dotenv"
dotenv.config()

import axios from "axios";
import fs from "fs/promises";
import cliProgress from "cli-progress";

const jsonpath = "list_of_repo.json";

const makeSearch = async (page: number) => {
  const config = {
    headers: {
      Authorization: `Token ${process.env.GITHUB_API_TOKEN}`,
    },
  };

  const url = `https://api.github.com/search/code?q=mongodb%2Bsrv+in:file&page=${page}&per_page=100`;
  const result: {
    items: { html_url: string; repository: { html_url: string } }[];
  } = await axios.get(url, config);

  // make an an object from result
  let obj = {};
  result.data.items.forEach((item) => {
    obj[item.repository.html_url] = item.html_url;
  });

  await addToJson(jsonpath, obj);
};

async function addToJson(jsonpath: string, data?: object) {
  const oldJson = (await fs.readFile(jsonpath)).toString();
  let jsonData = JSON.stringify(data, null, 2);

  if (oldJson) {
    jsonData = JSON.stringify(
      { ...JSON.parse(oldJson), ...JSON.parse(jsonData) },
      null,
      2
    );
  }

  await fs.writeFile(jsonpath, jsonData);
}

async function main() {
    // I included a CLI progress loader because, who doesn’t like a loader.
  const bar1 = new cliProgress.SingleBar(
    {},
    cliProgress.Presets.shades_classic
  );
  // number of iteration 10
  bar1.start(10, 0);
  for (let i = 1; i <= 10; i++) {
    await makeSearch(i);
    bar1.update(1);
  }
  bar1.stop();
}

main();

The results provided does not mean that an actual MongoDB connection string exists, it only implies that the repositories in the result have an in-file code that matches mongodb+srv:

I could go further to create a script to run a search through each code URL and run a regex to further find an actual connection string but that won’t be necessary as my purpose is to create public awareness and how to protect ourselves.

What I discovered and how we can protect ourselves

Some of my discoveries include:

• some of the results include old commits in the commit history: Just like my mistake that led to this article, sometimes we forget to create a .gitignore file at the beginning of a project and have some secrets staged somewhere in the commit history.

  We can make use of tools like GitGuardian to continually scan our repo for secrets in our source code.

• some results included messages from different log files and environment files: This probably happened due to not including a .gitignore.

  GitHub provides a repo with numerous type of .gitignore templates for different language, framework, tools, IDE e.t.c.

  And I created a simple interactive CLI to generate .gitignore templates based on the GitHub lists.

You can find the Interactive CLI tool to generate your .gitignore templates here: https://www.npmjs.com/package/gittyignore

Thanks for reading through! 🤗

You can buy me a coffee if you enjoy this article!