JFrog’s Advanced Security Scanners Discovered Thousands of Publicly Exposed API Tokens – And They’re Active

JFrog Advanced Security - Secrets Detection

Read our full research report onInfoWorld

The JFrog Security Research team released the findings of a recent investigation wherein they uncovered thousands of publicly exposed,activeAPI tokens. This was accomplished while the team tested the newSecrets Detectionfeature in the company’sJFrog Advanced Securitysolution, part ofJFrog Xray.

The team scanned more than eight million artifacts in the most common open-source software registries: npm, PyPI, RubyGems, crates.io & DockerHub (bothDockerfilesand small Docker layers). Each artifact was analyzed using the Secrets Detection scanners to find and verify leaked API tokens. For npm and PyPI packages, the scan also included multiple versions of the same package to try and find tokens that were once available but removed in a later version.

Pie chart displaying number of artifacts that were analyzed by JFrog Secrets Detection by platform. DockerHub made up the biggest slice, with 5.78 million of the 8 million scanned artifacts.
Analyzed artifacts per platform (in millions).


After scanning all supported token types by Secrets Detection and verifying the tokens dynamically, AWS, GCP, and Telegram, API tokens were the most leaked tokens (in that order). Interestingly, AWS developers seemed more vigilant about revoking unused tokens, with 47% of AWS tokens found active, unlike GCP, which boasts an active token rate of ~73%.

Graph displaying data of the distribution of active/inactive tokens for repositories. AWS, GCP, and Telegram API tokens were the most-leaked tokens.
Distribution of active/inactive tokens for each repository.

Although the initial goal of their research was to find and fix false positives using JFrog Advanced Security, the research team uncovered more active secrets than expected, which prompted the detailed analysis. To complete the analysis, the team privately disclosed all leaked secrets to their respective code owners (ones who could be identified), offering them a chance to replace or revoke the secrets as needed.

Secrets Detection uncovered disclosed secrets in the source code, like plaintext API keys, credentials, expired certificates, or passwords, often forgotten and left exposed unintentionally. These secrets threaten software’s integrity, allowing bad actors to access confidential information, data, or private networks.

深入阅读JFrog安全研究成果and the five best practices they recommend for safely storing tokens in thisInfoWorld article, and stay tuned for more examples of how the new features of JFrog Advanced Security can help safeguard your software supply chain.