**Title:** Sensitive Information Disclosure – Plaintext Credentials Exposed in /db/drug_recommendor.sql **Severity:** High **Accepted: CVE-2026-92927** **CWE:** CWE-200 – Exposure of Sensitive Information to an Unauthorized Actor **Researchers:** Karan Parelkar, Anubhav Verma, Parth Desai **Source / Vendor:** (Sourcecodester's Drug Recommendation System Using Machine Learning, PHP, and MySQL Database ) https://www.sourcecodester.com/php/18278/drug-recommender-web-app-student-project.html **Summary** During security testing of the open-source Drug Recommendor project, plaintext credentials were found in the publicly distributed database file: **/db/drug_recommendor.sql** The credentials are accessible by simply downloading and opening the SQL file, without authentication or exploitation. **Steps to Reproduce** 1. Navigate to /db/. ![POC](images/sql%20file%201.png) 2. Click on drug_recommender.sql ![POC](images/sql%20file%201.png) 3. Open drug_recommendor.sql in VS Code or any text editor. ![POC](images/sql%20file%202.png) 4. Credential information can be observed directly in the SQL dump. ![POC](images/sql%20file%203.png) **Impact** An attacker who obtains valid credentials may potentially gain unauthorized access to the associated application/account and sensitive information. **Remediation** Remove credentials from publicly distributed SQL dumps. Replace them with dummy/test credentials. Never store passwords in plaintext; use bcrypt/Argon2id. Rotate any credentials that have already been exposed. **Researchers Information** **Researcher - 1** Name: Karan Parelkar Independent Security Researcher Email: karan.parelkar2005@gmail.com GitHub: https://github.com/KaranParelkar LinkedIn: https://www.linkedin.com/in/karan-parelkar-6a370125b/ **Researcher - 2** Name: Anubhav Verma Independent Security Researcher Email: avdzav10@gmail.com GitHub: https://github.com/anubhavv106 LinkedIn: https://www.linkedin.com/in/anubhav-verma-7123a1232/ **Researcher - 3** Name: Parth Desai Independent Security Researcher Email: ppdesai3@asu.edu GitHub: https://github.com/ParthD31 LinkedIn: https://www.linkedin.com/in/parth-desai-801951224/