Arbitrary Code Injection Affecting flowise package, versions <3.1.3
Threat Intelligence
Snyk has a proof-of-concept or detailed explanation of how to exploit this vulnerability.
The probability is the direct output of the EPSS model, and conveys an overall sense of the threat of exploitation in the wild. The percentile measures the EPSS probability relative to all known EPSS scores. Note: This data is updated daily, relying on the latest available EPSS model version. Check out the EPSS documentation for more details.
Do your applications use this vulnerable package?
In a few clicks we can analyze your entire application and see what components are vulnerable in your application, and suggest you quick fixes.
Test your applicationsSnyk Learn
Learn about Arbitrary Code Injection vulnerabilities in an interactive lesson.
Start learning- Snyk IDSNYK-JS-FLOWISE-21693810
- published24 Oct 2026
- disclosed7 Oct 2026
- creditPeyton Kennedy
Introduced: 7 Oct 2026
NewCVE-2026-73487 (opens in a new tab)How to fix?
Upgrade flowise to version 3.1.3 or higher.
Overview
flowise is a Flowiseai Server
Affected versions of this package are vulnerable to Arbitrary Code Injection via validatePythonCodeForDataFrame() in packages/components/src/pythonCodeValidator.ts, which uses a regex-based blocklist to sanitize LLM-generated Python code before execution in Pyodide. An attacker can inject a prompt through the unauthenticated prediction API (POST /api/v1/prediction/:id) to steer the LLM into generating Python code that calls URL-accepting pandas functions such as pd.read_json(), pd.read_csv(), or pd.read_html() with an attacker-controlled URL, none of which match any of the 38 blocked patterns. This allows full exfiltration of loaded CSV or Airtable data to an external server, SSRF against internal services including cloud metadata endpoints, and potential further code execution via np.ctypeslib or chr()-based identifier construction. The supplier's own CVSS 3.1 vector (AV:N/AC:L/PR:N/UI:N/S:C/C:H/I:L/A:N, score 9.3) reflects the scoped, cross-boundary impact; the supplied record vectors differ from the source advisory's self-assessed vector, and we prefer the source advisory's vector as it is the most specific and best-evidenced assessment of the actual impact.