# CVE-2025-27772: UpTrain remote code execution in /new_run endpoint

Source: https://junglewise.ai/threats/cve-2025-27772-uptrain-remote-code-execution-in-new-run-endpoint  
Publisher: Junglewise Threat Intelligence, https://junglewise.ai/threats  
Published: 2026-08-17. Updated: 2026-09-25  
License: CC BY 4.0 with a link back

CVE-2025-27772 · Severity: info · CVSS 8.8 · Published 2026-08-17

UpTrain is an open-source platform for evaluating and improving generative AI applications. The `/new_run` endpoint fails to properly validate user input, allowing authenticated users to execute arbitrary code on the host system running UpTrain (typically a Docker container). This could allow an attacker with valid API credentials to take full control of the application infrastructure.

Technologies: [UpTrain](https://junglewise.ai/threats/technologies/uptrain). Vendors: [UpTrain](https://junglewise.ai/threats/vendors/uptrain).

## Quick answers

**Is CVE-2025-27772 exploited in the wild?** No. Junglewise Threat Intelligence has no report of CVE-2025-27772 being exploited in the wild.

**How severe is CVE-2025-27772?** Informational severity, with a CVSS score of 8.8, published 17 August 2026.

**Which products and versions are affected?** UpTrain 0.7.1 and prior.

**Is there a fix for CVE-2025-27772?** No fixed version is recorded yet. Check the vendor's advisory in the references below. If you run it, Junglewise drafts a mitigation that fits your setup and applies it once you approve, with senior security engineers behind it.

**How is CVE-2025-27772 attacked?** It is attacked over the network.

**Am I affected by CVE-2025-27772?** Check whether you run UpTrain at an affected version. Junglewise does that check for you around the clock: connect your code, cloud and domains, and it matches new vulnerabilities like this one against what you actually run and tells you when one reaches you.

## Technical details

The `/new_run` endpoint in UpTrain v0.7.1 and prior uses Python's `eval()` function to process the `checks` and `metadata` parameters without sanitization, leading to arbitrary code execution. The vulnerable code directly evaluates these form parameters: `checks = eval(checks[0])` and `metadata = eval(metadata)`. Exploitation requires a valid API key (authenticated access) and network access to the UpTrain dashboard. An attacker can craft malicious Python expressions in the `checks` or `metadata` fields to execute arbitrary code in the context of the running process. As of publication, no patch is available; the project appears to be unmaintained.

## Affected products

- UpTrain UpTrain 0.7.1 and prior

## Timeline

- 2024-09-05: disclosed: Issue opened in repository
- 2025-03-10: other: Project indicated to be unmaintained via Slack messages
- 2026-08-10: other: Fix attempt via pull request (status unknown)
- 2026-08-17: advisory: Public advisory published

## References

- [https://github.com/uptrain-ai/uptrain/blob/a31cc14eddcb6c0b0b12cbed15f086d98c441c6f/uptrain/dashboard/backend/app.py](https://github.com/uptrain-ai/uptrain/blob/a31cc14eddcb6c0b0b12cbed15f086d98c441c6f/uptrain/dashboard/backend/app.py)

## Related threats

- [UpTrain remote code execution in /add_prompts endpoint](https://junglewise.ai/threats/cve-2025-27771-uptrain-remote-code-execution-in-add-prompts-endpoint)
- [UpTrain remote code execution in /create_project endpoint](https://junglewise.ai/threats/cve-2025-27770-uptrain-remote-code-execution-in-create-project-endpoint)