Executive brief
FlowiseAI, a platform for building LLM applications, contains a vulnerability that allows authenticated users to view encrypted credentials such as API keys and passwords. If an attacker also gains access to the system's encryption key file, they can fully decrypt and steal sensitive service tokens for AWS, OpenAI, and other integrated services. This could lead to unauthorized access to third-party cloud accounts and data breaches.
Technical details
An information disclosure vulnerability (CWE-200) exists in FlowiseAI's credential management service. The root cause is located in `packages/server/src/services/credentials/index.ts`, where the application fails to strip the `encryptedData` field from the database response when a `credentialName` filter is applied to the API request. While the application correctly omits this field during unfiltered requests, the filtered code path directly pushes the database objects into the response array. An authenticated attacker can exploit this by sending a crafted GET request to `/api/v1/credentials` with a filter parameter. If the attacker has read access to the default encryption key file (`~/.flowise/encryption.key`), they can decrypt the leaked AES-encrypted strings to obtain plaintext API keys and tokens. The issue is fixed in version 3.1.2.
Affected products
- FlowiseAI flowise <= 3.1.1
Timeline
- 2026-05-14: disclosed
- 2026-05-14: advisory
- 2026-05-14: patched: Fixed in version 3.1.2