Executive brief
Flowise is a low-code platform for building conversational AI applications. The vulnerability allows authenticated users to escape the JavaScript sandbox and execute arbitrary commands on the server (as root in Docker deployments) or read any file on the host system. This can lead to complete system compromise and exposure of sensitive data.
Technical details
The vulnerability is a sandbox escape in the vm2/@flowiseai/nodevm JavaScript sandbox used to safely execute custom user code. The root cause is the inclusion of 'puppeteer' and 'playwright' in the allowlist of available dependencies. An authenticated user with access to the /api/v1/node-custom-function endpoint can call puppeteer.launch() with attacker-controlled executablePath and args parameters, which internally invokes child_process.spawn() outside the sandbox boundary. In versions 2.0.0–3.0.7, this is exploitable by default; in 3.0.8–3.1.2, exploitation requires the ALLOW_BUILTIN_DEP=true environment variable. Secondary exploitation via puppeteer's headless mode allows arbitrary file read on the host (e.g., via file:// URLs). The vulnerability was patched in version 3.1.3.
Affected products
- Flowise flowise <= 3.1.2
- Flowise flowise-components <= 3.1.2
Timeline
- 2026-07-29: disclosed: GitHub Security Advisory GHSA-9gvv-qjj3-2p6g published
- 2026-08-13: patched: Fixed in version 3.1.3
- 2026-08-13: advisory: CVE-2026-73483 published