Summary
Flowise contains a missing authorization vulnerability in the OpenAI assistants file download endpoint, which is exposed through the global authentication whitelist. An unauthenticated attacker who provides valid chatflow, chat, and file identifiers may retrieve private files from other chatflows, workspaces, or organizations.
Affected context
The issue affects Flowise deployments using versions 2.2.4 through 3.1.4 when the
POST /api/v1/openai-assistants-file/download endpoint is included in the global
authentication whitelist. In this configuration, requests may bypass session and API
key verification while still reaching file retrieval logic for chatflows that the
requester does not own or otherwise have permission to access.
Exploit preconditions
- Attack surface: The OpenAI assistants file download API endpoint
- Required access: No authenticated session or API key may be required when the endpoint is whitelisted
- Trigger point: A file download request containing valid
chatflowId,chatId, andfileNameidentifiers - Why exploitation is possible: The endpoint can process file retrieval without enforcing authorization against the requested chatflow, workspace, organization, or file
Why this happens
The vulnerability results from an authorization control gap in which an endpoint is
globally exempted from authentication without applying an equivalent resource-level
authorization check. This maps to CWE-862: Missing Authorization, because the
application may verify neither the caller's identity nor the caller's permission to
access the referenced private resource before returning its contents.
Potential impact
- An unauthenticated attacker may retrieve private files
- Confidential data from other chatflows may be disclosed
- Files belonging to separate workspaces or organizations could be exposed
- Exposed files may support further account, application, or data compromise
General mitigation guidance
Mitigation generally focuses on enforcing authorization for every file access request,
including requests that use identifiers supplied by the client.
Common defensive patterns include:
- Avoiding broad authentication whitelists for resource-serving endpoints
- Requiring authenticated identity and API authorization before file retrieval
- Checking ownership, workspace membership, organization scope, and resource permissions
- Binding file access to server-side chatflow and chat records rather than trusting identifiers alone
- Returning uniform errors that do not disclose whether private resources exist
- Applying least-privilege storage permissions and isolating tenant data
How teams detect issues like this
Organizations can look for:
- Static analysis rules identifying endpoints excluded from authentication middleware
- Code review signals where file access relies on client-supplied resource identifiers
- Tests that exercise cross-user, cross-workspace, and unauthenticated access attempts
- API gateway or policy rules requiring authentication for file download routes
- Audit logs showing file downloads without a corresponding authenticated principal or authorization decision
Automated security testing and code review tools may help identify inconsistent authentication and authorization coverage.

