Payload CMS: SQL injection in SQLite and Postgres adapters via dynamic filters and joins

Summary

Payload is a headless content management system. Versions 3.0.0 before 3.88.0 and 4.0.0-canary before 4.0.0-canary.27 contain an SQL injection flaw in the SQLite and Postgres database adapters. An untrusted user who can query readable collections through dynamic filters or joins can craft a request that injects SQL, which may allow reading, changing or deleting database content. MongoDB installations are not affected.

Affected

• Payload (npm package payload) 3.0.0 up to but excluding 3.88.0
• Payload 4.0.0-canary.0 up to but excluding 4.0.0-canary.27
• Only SQLite and PostgreSQL adapters; MongoDB is not affected

Fixed in

• Payload 3.88.0
• Payload 4.0.0-canary.27

Workaround

• No vendor workaround. The vendor advises upgrading.
• Temporary only: restrict access to affected API paths and review database logs for unusual query errors.

Exploitation

• CVSS 3.1 score 9.8 (critical), network attack vector, low complexity, no privileges and no user interaction required.
• CISA SSVC: automatable yes, technical impact total, exploitation none.
• Not marked as known exploited at the time of research. Public proof of concept: unknown.

Our recommendation

• Inventory all Payload instances and check the payload and @payloadcms/* package versions in lockfiles.
• Check whether the instance uses SQLite or PostgreSQL and whether any collection is publicly readable with filter or join support.
• Upgrade to 3.88.0, 4.0.0-canary.27 or later.
• If patching is delayed, restrict affected API paths and monitor database logs.

Public exploits iAttack code is publicly available (PoC on GitHub, Metasploit or Exploit-DB).

PoC-in-GitHub, Metasploit and Exploit-DB list nothing for this CVE.

Research sources

• Vendor advisory GHSA-v49j-62m6-pgrr github.com
• Fix commit a742140 github.com
• Release v3.88.0 github.com
• GitHub Advisory Database GHSA-v49j-62m6-pgrr github.com
• OSV CVE-2026-105845 osv.dev

Researched and written up in English and German by an AI (Claude by Anthropic) based on NVD, CISA, EPSS and the sources listed. The vendor information is authoritative. Researched on 30 Oct 2026, 21:06.