The number of cyberattacks is growing rapidly and there are too many software vulnerabilities to fix them all. Researchers at Leiden University have developed a new method to help identify which vulnerabilities deserve attention first: the GitHub Popularity Metric.
In the first ten months of 2026, more than 82,000 software vulnerabilities have been reported. That is more than double the number reported during the same period in 2025. Advanced artificial intelligence makes it easier to discover weaknesses faster.
'Defenders must protect their systems against many vulnerabilities, while an attacker may need to exploit just one.'
For organisations and their IT experts, that creates a difficult situation. These vulnerabilities need to be fixed, but time and resources are limited. 'Attackers have a significant advantage. Defenders must protect their systems against many vulnerabilities, while an attacker may need to exploit just one', says PhD candidate Jafar Akhoundali from the Leiden Institute of Advanced Computer Science (LIACS).
Which vulnerabilities should be fixed first?
Prioritising is one of the solutions. Security teams fix those vulnerabilities that are more likely to be exploited first. There are several systems designed to help prioritisation. One is the straightforward Common Vulnerability Scoring System (CVSS), which rates the severity of a vulnerability from 0 to 10. The Exploit Prediction Scoring System (EPSS) estimates the likelihood that a known software vulnerability, listed in the Common Vulnerabilities and Exposures (CVE) database, will be exploited by attackers in the wild within the next 30 days. Another useful resource is the CISA Known Exploited Vulnerabilities (KEV) catalog, which lists CVEs that have been actively exploited.
Measuring the popularity of exploit code
Leiden computer scientists have now added a new approach: the GitHub Popularity Metric (GPM). GitHub is a platform where developers share software source code. This includes proof-of-concept code that demonstrates how security vulnerabilities can be exploited. Akhoundali, together with Kristian Rietveld and Olga Gadyatskaya, investigated whether the attention a proof-of-concept code receives, could help identify vulnerabilities that are more likely to be exploited.
They analysed nearly 20,000 GitHub repositories associated with more than 8,000 vulnerabilities. Vulnerabilities attracting greater attention were found to be more frequently associated with known real-world exploitation.
Detecting overlooked vulnerabilities
The researchers identified CVEs that ranked low on other risk lists because they hadn't been exploited yet but attracted considerable attention on GitHub. In some cases, high GitHub popularity signalled early interest in a vulnerability, even before its official publication. The team identified a vulnerability that had officially been marked as rejected, suggesting it was not a real security issue. Yet, its unusually high GitHub popularity attracted attention. Shortly afterwards, leaked communications from the Black Basta ransomware group referred to the same vulnerability. Its official status was subsequently changed to published (actually recognised as a vulnerability).
'We provide is a new tool in the toolbox.'
A new powerfool tool
'Each existing system has its advantages,' says Akhoundali. 'Ours too, because we measure a previously underexplored factor: the popularity of proof-of-concept code, using publicly available data. But inevitably there will be exploited vulnerabilities that are not yet popular on GitHub. It's a new tool in the toolbox.' That is why Akhoundali is happy to see the Leiden GPM incorporated into version 5 of EPSS. 'Our method can be used independently, but it is more useful if it's integrated.'
The LIACS team will be demonstrating their metric at the ACM Conference on Computer and Communications Security (CCS 2026) in The Hague from 15 to 19 November. 'We will present visualisations and time-based analyses showing how the GitHub Popularity Metric can be used to assess CVE's,' says Akhoundali. 'We will also make the source code of the GPM publicly available.'