Servers and applications
CVE-2026-91935
SSRF + LLM API key exfiltration via user-controlled basepath / baseUrl in 8 chat-model nodes
- Severity
- Critical 9.4
- Published
- 31 Aug 2026
- Last changed
- 20 Sept 2026
- Added to UpHeron
- 1 Nov 2026
- Fixed in
- 3.1.4
- Exploit chance
- 0.4% in the next 30 daysHigher than 33% of CVEs. EPSS by FIRST, 19 Nov 2026
- Weakness
- Information exposure (CWE-200)Insufficiently protected credentials (CWE-522)Server-side request forgery (CWE-918)
- Attack
- Over the network · Needs a user account · No user action needed
- Attacker can
- Read data · Change data · Slow it down
- CVSS vector
CVSS:4.0/AV:N/AC:L/AT:N/PR:L/UI:N/VC:H/VI:H/VA:L/SC:H/SI:H/SA:L- Also known as
- GHSA-hx55-h48h-7rw9
- Sources
- Project advisories (GitHub)
Description
Flowise chat-model nodes accept an attacker-controlled basepath / baseUrl input and pass it directly into the underlying LangChain SDK as the LLM provider's baseURL, bypassing Flowise's own packages/components/src/httpSecurity.ts SSRF-protection module. The HTTP Request node, Firecrawl, Searxng, RequestsGet, and ~35 other outbound-HTTP call-sites correctly use secureAxiosRequest(...) / secureFetch(...) — which enforce a deny-list covering 127.0.0.1, 169.254.169.254 (AWS IMDS), RFC1918 ranges, link-local, and IPv6 loopback. Eight chat-model nodes do not: ChatOpenAICustom, ChatOllama (both variants), ChatLocalAI, ChatLitellm, ChatNvdiaNIM, ChatNemoGuardrails, ChatCloudflareWorkersAI, and ChatGoogleGenerativeAI. Each of them hands the user-supplied URL straight to LangChain's provider SDK, which then calls fetch() against that URL server-side with no Flowise-side validation.
Affected software
| Software | Affected versions |
|---|---|
FlowiseServers and applications · github:flowiseai/flowise |
|
About Flowise
Build AI Agents, Visually
- Archived
- Yes, no longer maintainedGitHub
- Weekly downloads (npm)
- 6,604npm, 15 Nov 2026
- Docker Hub pulls
- 7,390,661Docker Hub, 15 Nov 2026
- GitHub stars
- 55,573GitHub, 15 Nov 2026
- Forks
- 25,094GitHub, 15 Nov 2026
- Latest release
- [email protected]GitHub, 15 Nov 2026
- Released
- 29 Jul 2026GitHub
- First published on GitHub
- 31 Mar 2023GitHub
Source code Website Docker image Package
Vulnerabilities in Flowise
- In our list
- 171
- Published in the last 12 months
- 149
- Critical, last 12 months
- 41
- High, last 12 months
- 70
- Known exploited
- 0
- CVE-2026-130796Flowise 3.1.3 through 3.1.4 MCP Server Token Disclosure via Public Chatflow EndpointHigh 8.8 17 Nov 2026
- CVE-2026-130615Flowise through 3.1.4 SSRF via DNS Rebinding in Playwright and Puppeteer LoadersHigh 7.6 17 Nov 2026
- CVE-2026-127214Flowise 3.1.4 SSRF Bypass via Custom Function SandboxMedium 6.3 13 Nov 2026
- CVE-2026-119978Flowise through 3.1.4 Custom Function Sandbox Escape RCE via LocalFileStoreCritical 9.0 30 Oct 2026
- CVE-2026-90580A vulnerability was found in FlowiseAI Flowise up to 3.0.2. This vulnerability affects the function axios.post of the file packages/server/src/controllers/evaluations/index.ts of the component Evaluations Endpoint. The manipulation of the argument Host/X-Forwarded-Proto results in server-side request forgery. The attack may be launched remotely. The exploit has been made public and could be…Low 2.1 13 Sept 2026
- CVE-2026-90535Flowise versions before 3.1.4 contain an unauthenticated denial of service vulnerability in the /api/v1/text-to-speech/abort endpoint that accepts user-supplied chatflowId and chatId without ownership verification. Attackers can terminate active chatflow predictions for any user by submitting requests with known chatflow and chat identifiers, causing targeted service disruption.Medium 6.3 12 Sept 2026
References
- CVE.org https://www.cve.org/CVERecord?id=CVE-2026-91935
- GitHub Advisory Database https://github.com/advisories/GHSA-hx55-h48h-7rw9
- github.com https://github.com/FlowiseAI/Flowise/security/advisories/GHSA-hx55-h48h-7rw9