Executive brief
Flowise, a low-code platform for building AI workflows, has a vulnerability in its credential management API that exposes stored secrets in plaintext. When authenticated users retrieve stored credentials (such as database passwords, API keys, and cloud service account files) via the GET /api/v1/credentials API endpoint, sensitive data stored in string-type fields bypasses the redaction filter and is returned unmasked, allowing any user with credentials:view permission to access production secrets like MongoDB passwords, AWS keys, and Google service account private keys.
Technical details
The vulnerability is an incomplete credential redaction flaw in the credential retrieval endpoint. The GET /api/v1/credentials/:id endpoint decrypts stored credential data and returns it in the plainDataObj response field. While a redactCredentialWithPasswordType() function exists to mask sensitive fields, it only redacts fields explicitly typed as 'password' in the component schema. Many credential types store secrets in fields defined as type: 'string' (e.g., MongoDB connection URLs with embedded passwords, Google service account JSON with RSA private keys, AWS access keys, PostgreSQL URLs), which are returned in plaintext without any masking. An authenticated attacker with credentials:view permission can retrieve the raw unencrypted secrets of any credential in the workspace. The patch (version 3.1.3+) extends redaction to cover all sensitive credential fields regardless of their schema type.
Affected products
- FlowiseAI Flowise <= 3.1.2
Timeline
- 2026-07-27: disclosed
- 2026-08-04: advisory
- 2026-08-05: patched: Fixed in version 3.1.3