Executive brief
Flowise, an open-source tool for building LLM apps, contains a vulnerability where sensitive credentials like database passwords and cloud access keys are exposed in plain text through its API. While the system attempts to hide passwords, it fails to redact sensitive information stored in standard text fields, such as database connection strings or service account keys. An authorized user could exploit this to steal credentials for connected services like AWS, MongoDB, or Google Cloud, potentially leading to unauthorized access to corporate data and infrastructure.
Technical details
An information disclosure vulnerability exists in the `GET /api/v1/credentials/:id` endpoint of Flowise due to incomplete redaction logic in the `redactCredentialWithPasswordType()` function. The application only masks fields explicitly defined with `type: 'password'` in the component schema, leaving sensitive data stored in `type: 'string'` fields (such as MongoDB connection URLs, AWS keys, and Google Service Account JSONs) exposed in the `plainDataObj` field of the API response. An authenticated attacker with `credentials:view` permissions can retrieve these decrypted secrets in plaintext. The issue is addressed in version 3.1.3 by improving redaction coverage.
Affected products
- FlowiseAI flowise <= 3.1.2
Timeline
- 2026-07-27: disclosed
- 2026-07-27: patched: Fixed in version 3.1.3
- 2026-08-04: advisory