French cybersecurity company CrowdSec has revealed that in a May supply-chain attack tied to compromised TanStack npm packages, approximately 170 of its private GitHub repositories were copied using access belonging to a former employee. The breach also exposed email addresses belonging to users and investment details from potential backers.
What Happened in the Attack
On May 11, attackers published 84 malicious versions of dozens of TanStack npm packages. These trojanized packages, later identified under CVE-2026-45321, executed code that harvested credentials such as GitHub tokens, SSH keys, and cloud credentials from infected developer machines. Through one compromised account—specifically, that of an employee who had recently resigned but whose account access had been retained—attackers gained entry and made the code copy on May 22.
CrowdSec terminated the employee’s remaining access on May 25. However, by then the damage was done: the copy was made with an OAuth token kept from that user’s account, which was leveraged to access the company’s private GitHub organization. CrowdSec maintains that its infrastructure and databases, including AWS systems, were not infiltrated.
What Was Exposed
CrowdSec says the leak surfaced on September 16. It included: source code from internal, non-public repositories; email addresses of 83 users; and names, emails, and investment contexts from 51 potential investors dating back to 2020, drawn from a system never meant for public access. The code stolen covers critical internal tools—its web console, data science models, automation scripts, and the consensus algorithm that powers how IP addresses get chosen for its blocklists.
The algorithm’s thresholds—how many detections it needs before blacklisting an IP—were among the newly exposed secrets. CrowdSec stresses the leaked versions are nearly four months old and that many internal credentials had already been rotated or were otherwise unusable. The only credential that may have remained usable was an AWS SNS notification service token that was tested on August 17, though unsuccessfully.
Similar Fallout at Other Companies
Other organizations were impacted by the same TanStack npm compromise. Developer devices at Mistral AI were infected, and at OpenAI two devices were affected, leading to limited access of internal code repositories. The broader campaign highlights how such supply-chain attacks can ripple across trusted ecosystems.
CrowdSec’s Response and What’s Changing
In its initial statement (released before September 18), CrowdSec claimed no user or investor data had leaked, asserting the impact was confined to internal code. By the comprehensive September 18 report, however, the company acknowledged that personal details were indeed exposed, including those of users and investors. CEO Philippe Humeau issued a personal apology to investors.
Following the exposure, credentials found in the leak were rotated around September 16-17. CrowdSec has also since mandated endpoint protection software on all staff machines working with internal systems or code. While the company says it found no malicious TanStack versions in its codebase, it now believes the former employee’s retained GitHub token was the vector.
What this means: the incident underscores the severe downstream risk posed by supply-chain vulnerabilities—especially when combined with the retention of access for departed employees. Organizations need to monitor all dependencies, promptly revoke access when no longer required, and treat token security as critically as any production credential. The exposed thresholds for CrowdSec’s blocklist algorithm may increase the likelihood of abuse or manipulation, so watch how they respond. If they change their thresholds or structure their detection consensus, the trust model they offer—and that’s part of many security-vendors’ appeal—could shift significantly.
