CVE-2026-3057 is a medium severity flaw (CVSS 6.3) in pearProjectApi. The projectCode argument in the task model's date total function reaches the database unsafely, allowing SQL injection. It can be attempted remotely by an authenticated user, and the exploit is public.
What's affected
The NVD record lists pearProjectApi up to and including version 2.8.10. The NVD record does not name a fixed version, so assume every copy you run is exposed until the maintainer says otherwise. The vendor was contacted early about this disclosure but did not respond, according to the record. The public disclosure and the record are the only sources for this issue.
How to check if you're exposed
Find out whether pearProjectApi runs anywhere in your environment, including test servers and old demos. Check the version in your deployment against 2.8.10, and look at the backend project files for application/common/Model/Task.php.
Then check who can reach it. A copy that is open to the internet is the case that matters most, so review the web server, reverse proxy and firewall rules for that host.
What to do now
- Remove pearProjectApi from any internet-facing host, or restrict it to a trusted network or VPN.
- Check the project page and the maintainer's contact channels for a fix, and apply it if one has appeared since this was written.
- Review who holds accounts, since the vector needs a low privilege login, and remove accounts you do not need.
- Back up the database and review project and task data for unexpected changes.
- If you maintain a fork, use parameterised queries in the task model.
- Keep a list of small third party applications like this one, so a new advisory can be matched against it quickly. The Gold page for this CVE has the latest data: https://gold.safeguard.sh/vulnerability/CVE-2026-3057
Why it matters
The record files this under CWE-74 and CWE-89. The input is placed into a database query without being neutralised, which is the textbook form of CWE-89. A request parameter that should hold a plain value is read as part of the query itself, so the attacker can change what the database does. Parameterised queries are the standard fix, and they are a change the maintainer or your own developers must make in the code.
The NVD description says the exploit is public, so anyone can read how the weakness works. A public write-up and no vendor fix is a poor mix for a system that holds real data, which is why exposure control is the main defence here. The record does not mark this CVE as exploited in the wild, and it is not in the CISA KEV catalog.
Quick check
Does this need a login?
The CVSS vector says low privileges, so an ordinary user account is enough.
Did the vendor respond?
According to the record, the vendor was contacted early and did not respond.
Sources
Last reviewed: 19 October 2026.
