Executive brief
OpenHarness, an open-source AI development framework, contains a security flaw in its chat-based command system. A remote user with chat access can use a specific command to bypass security restrictions and read sensitive files from the server hosting the application. This could lead to the exposure of private configuration files, system data, or other sensitive information stored on the host machine.
Technical details
A path traversal vulnerability exists in OpenHarness due to insufficient input validation in the '/memory show' slash command. The application joins attacker-controlled path parameters with the project memory directory without verifying that the resulting path remains within the intended directory boundaries. A remote attacker with chat access can provide traversal sequences (e.g., ../../) to escape the project memory directory and read arbitrary files accessible to the OpenHarness process. The issue was addressed in commit dd1d235 by enforcing filesystem containment and hardening the gateway's slash command security model.
Affected products
- HKUDS OpenHarness Prior to commit dd1d235450dd987b20bff01b7bfb02fe8620a0af
Timeline
- 2026-04-12: disclosed: Initial pull request submitted by researcher
- 2026-04-13: patched: Fix merged into main branch in commit dd1d235
- 2026-04-16: advisory: CVE published to NVD