Is my AI tool safe?Agent frameworks and buildersFlowiseAI
Is Flowise safe?
No tool is safe or unsafe on its own: it depends on what it can reach and how it is set up. The record for Flowise: 1 incident in this center's library and 4 CVEs in its advisory feed (4 critical), the newest published Jul 29, 2026. Below: what went wrong, what was fixed, and how to use it safely.
What to watch for
Same shape as Langflow: a visual builder with critical CVEs that run code on the server, often deployed with a public address because that is how the tutorials show it.
How to use it safely
Authentication on, internet off, current version, and a separate secret store for the keys the flows use.
What it can reach
- Often exposed to the internet
- Runs commands with your permissions
- Holds API keys and tokens
What went wrong: incidents in the library
- MCP STDIO design flaw: server configs become command execution Critical
OX Security found that MCP's STDIO transport runs whatever command a server configuration names. Any AI app that lets users, web requests or an agent change that configuration can be turned into remote code execution, and the…
Known CVEs
- CVE-2026-73487 Critical Flowise CSV and Airtable agents run LLM-written Python past a bypassable blocklist
- CVE-2026-73483 Critical Flowise NodeVM sandbox escape through Puppeteer launch options
- CVE-2026-40933 Critical Flowise authenticated RCE via MCP adapters (allow-list bypass)
- CVE-2025-59528 Critical Flowise CustomMCP node remote code execution
How many people use it
- 55k GitHub stars · GitHub
- 2.9k npm downloads in a week (Sep 20 to 26, 2026) · npm download counts
A high CVE count can mean heavy scrutiny as much as weak engineering, so read the incidents and advisories, not only the number. The counts are what this center has recorded, not every flaw ever found; usage figures are the vendors' and registries' own published numbers.