A large-scale security audit of Bitcoin ecosystem codebases has revealed alarming results. Over 27.5 hours of continuous work, a team of specialists led by Cashu developer under the pseudonym calle discovered 4,962 vulnerabilities across 390 projects. Of these, 85 were deemed critical, and another 635 are classified as high severity.
How the audit was conducted
During the marathon, which was run around the clock by a globally distributed group of 16 people, both manual methods and automated AI-powered tools were used. Each participant configured their own AI agents, which allowed for a high diversification of findings. The average speed was 166.3 detected issues per hour, covering 13.1 projects in the same period.
Interestingly, a significant spike in data was linked to the upload of a large batch of checks by one participant named Rob. Financial and computational support for the project was provided by OpenSats, Finn Puklowski, Steve, as well as companies OpenCode, Wafer Ai, and Kimi_Moonshot.
Numbers and categories
Of the entire body of findings, 14.3% are of high or critical severity. The "medium level" accounts for 1,386 cases (28.1%), "low" — 1,723 (34.9%), and "informational" — 869 (17.6%). Only eight findings were classified as false positives.
The most vulnerable turned out to be cryptographic libraries and SDKs — 1,385 findings across 128 audited projects. Software wallets took second place with 742 findings, and infrastructure and tooling — 590. The smallest coverage was in the categories of research and documentation (12 findings) and privacy with mixing (50 findings).
Reaction and prospects
So far, only 141 messages have been delivered to project owners — just 2.9% of all findings. Dynamic reproduction was achieved for 21.4% of issues, with 91% of findings coming through automated scanning. Developers have already confirmed most critical vulnerabilities, and a significant portion of results is verified on a local regtest test environment before sending.
My expert opinion: Such initiatives highlight that even in mature crypto ecosystems, security remains the "Achilles' heel." The speed at which AI finds vulnerabilities suggests that traditional audits are outdated. The market needs new standards of proactive protection, otherwise we risk seeing a wave of exploits capable of undermining trust in Bitcoin as a reliable store of value.