Budibase SSRF in AI table generation via uploadUrl

Executive brief

Budibase is a low-code platform used to build business applications. A vulnerability in its AI table generation feature allows users with 'builder' permissions to trick the server into making requests to internal network resources. This could lead to the exposure of sensitive cloud metadata, such as security credentials, or allow an attacker to probe private internal services that are not normally accessible from the internet.

Technical details

A Server-Side Request Forgery (SSRF) vulnerability exists in Budibase's AI table generation feature. The 'uploadUrl()' function in 'fileUtils.ts' utilizes a bare 'fetch()' call without implementing the 'fetchWithBlacklist()' protection used elsewhere in the codebase. An authenticated user with builder-level privileges can craft prompts that cause the LLM to generate internal IP addresses or cloud metadata endpoints (e.g., 169.254.169.254) as attachment URLs. When the server processes these attachments, it fetches the content from the internal resource and saves it to object storage, allowing the attacker to exfiltrate sensitive data like AWS IAM credentials or scan internal network ports. The issue is addressed in version 3.39.4.

Affected products

• Budibase Budibase <= 3.39.0

Timeline

• 2026-07-22: disclosed
• 2026-07-24: advisory

References

• https://api.github.com/users/oduoke567
• https://github.com/oduoke567
• https://api.github.com/users/oduoke567/gists%7B/gist_id%7D
• https://api.github.com/users/oduoke567/repos
• https://avatars.githubusercontent.com/u/102439013?v=4
• https://api.github.com/users/oduoke567/events%7B/privacy%7D