Platform
python
Component
foundationagents
Fixed in
0.8.1
0.8.2
A code injection vulnerability has been identified in FoundationAgents MetaGPT, specifically affecting versions 0.8.0 through 0.8.1. This flaw resides within the check_solution function of the HumanEvalBenchmark/MBPPBenchmark component, enabling remote attackers to inject malicious code. The vulnerability is considered HIGH severity and a public exploit is now available, highlighting the urgency of remediation.
Successful exploitation of CVE-2026-5970 allows an attacker to execute arbitrary code on the system running FoundationAgents MetaGPT. This could lead to complete system compromise, data exfiltration, or denial of service. Given the remote nature of the attack and the availability of a public exploit, the potential for widespread exploitation is significant. The impact is amplified by the fact that MetaGPT is likely used in environments where sensitive data or critical processes are managed, potentially leading to cascading failures and significant reputational damage.
This vulnerability is actively being exploited, as evidenced by the public availability of a proof-of-concept. It was reported via a pull request, but the project has not yet responded. The vulnerability is not currently listed on KEV, and the EPSS score is likely to be assessed as medium to high due to the public exploit and ease of exploitation.
Exploit Status
EPSS
0.07% (21% percentile)
CISA SSVC
CVSS Vector
The primary mitigation for CVE-2026-5970 is to upgrade to a patched version of FoundationAgents MetaGPT. As of this writing, no patched version has been released. Until a patch is available, consider implementing input validation on the check_solution function to sanitize user-provided data. While not a complete solution, this can reduce the attack surface. Monitor system logs for suspicious activity related to the HumanEvalBenchmark/MBPPBenchmark component. Implement network segmentation to limit the potential blast radius of a successful attack.
The code injection (Code Injection) vulnerability in the `check_solution` function of MetaGPT can be mitigated by updating to a patched version. Since the project has not responded, it is recommended to review the affected source code and apply security patches manually or avoid using the vulnerable function until an official update is released.
Vulnerability analysis and critical alerts directly to your inbox.
CVE-2026-5970 is a code injection vulnerability affecting FoundationAgents MetaGPT versions 0.8.0–0.8.1. The check_solution function allows remote attackers to inject code, potentially leading to system compromise.
If you are using FoundationAgents MetaGPT versions 0.8.0 or 0.8.1, you are potentially affected by this vulnerability. Immediate action is required.
The recommended fix is to upgrade to a patched version of FoundationAgents MetaGPT. As of now, no patch is available. Implement input validation as a temporary mitigation.
Yes, a public exploit for CVE-2026-5970 is available, indicating active exploitation is likely occurring.
Check the FoundationAgents project repository and website for updates and advisories regarding CVE-2026-5970. As of this writing, no official advisory has been published.
Upload your dependency file and we'll tell you instantly if this and other CVEs hit you.
Upload your requirements.txt file and we'll tell you instantly if you're affected.