JFrog's lab found 42 hijacked repos wired to fire in Claude and Cursor; its stock is up 61% in 2026
JFrog's lab published three finds this week that all sit on AI tooling: credential-stealing LiteLLM wheels, 42 booby-trapped repos wired to Claude and Cursor, and an actively exploited Brotli flaw. The market has priced the fear; the November guide will show whether it converts to bookings.
Richard Tang · 2 min read
Three JFrog+0.98% — JFrog, up 0.98 percent today research finds in one week all sit on the same surface: the tooling developers now run AI with. Malicious LiteLLM 1.107.0 wheels on PyPI ran at Python startup and stole AI provider credentials, GitHub−0.80% — GitHub, down 0.80 percent today tokens and cloud keys 1. Forty-two hijacked GitHub repositories hid a 13 MB payload wired to fire the moment a developer opens a project in Claude, Gemini, Cursor or VS Code 23.
And Brotli2Shell, a CVSS 9.2 flaw in Google's Brotli decoder, is under active exploitation and forced emergency patches across Node.js, .NET and Python 45. The chain is clear: AI coding tools expanded what a developer's machine trusts, and attackers followed. Enterprises that put models and AI agents into their build pipelines now need to scan skills, plugins, MCP servers and AI packages the way they scan code. JFrog's lab finds these threats, and its AI Security product just extended scanning onto developer workstations, with customers running AI asset scans on their own model keys 6.
The market has priced the fear
KeyBanc reiterated JFrog Overweight on October 27, citing security demand, and the stock has returned roughly 61% this year against the S&P 500's 13.5% 7. JFrog's security core and cloud revenue drove its fiscal Q2 beat, with cloud revenue up 53% year on year 8.
JFrog has returned 60.8% this year, triple the S&P 500
Data
| Value | |
|---|---|
| JFrog (FROG) | +60.8% YTD |
| QQQ | +17.7% YTD |
| S&P 500 | +13.5% YTD |
The November print is the test
Management guides November revenue to $164 million to $166 million, about 20% growth, down from 28.7% in fiscal Q2, and keeps forecasts cautious on usage above customers' contractual commitments 8. The finds published this week are exactly what the newest product line scans for. The November report is the first test of whether the lab's fear converts into bookings, or only into followers.
The growth engine slows: cloud 53%, total 28.7%, the November guide about 20%
- Year-on-year growth
- Estimate
Data
| Year-on-year growth | |
|---|---|
| Fiscal Q2 2026 cloud revenue | +53% |
| Fiscal Q2 2026 total revenue | +28.7% |
| November guide, total revenue (estimate) | +20% |
For teams, the takeaway stands on its own: verify the wheel you install, not just the source, and treat AI tool config files as attack surface. Start there, and learn what your own AI stack actually trusts.
Get Security by email, free
The breaches, flaws and security vendors AI is putting to the test. In our daily email, every weekday.
