CVE-2026-114106 HKUDS CVE debrief

Who should care

Defenders responsible for deploying and managing the Vibe-Trading AI tool should assess exposure and prioritize patching to prevent unauthorized access to sensitive files. This includes reviewing container configurations, restricting access to the LLM tool registry, and monitoring for suspicious activity.

Why it matters

The CVE-2026-114106 vulnerability in the Vibe-Trading AI tool allows for arbitrary file reads due to flawed `safe_user_path()` and `read_document()` functions. Defenders should prioritize patching to prevent unauthorized access to sensitive files. The vulnerability affects the auto-discovered LLM tool registry and can be exploited by anonymous TCP clients to port 8899. Evidence is based on official CVE and source records.

• Potential unauthorized access to sensitive files
• Possible exposure of secret files mounted into the container
• Risk of reading system files like `/etc/shadow` and `/etc/passwd`
• Need for verification of patching and fix effectiveness

Technical summary

The `safe_user_path()` function in the Vibe-Trading AI tool accepts any path under `Path.home()` or `Path.cwd()`, which can resolve to sensitive areas like `/root` and `/app` in the container. The `read_document()` function lacks a sandbox call and can return the full content of any path readable by the FastAPI process, including sensitive files like `/etc/shadow` and `/etc/passwd`. The container runs as `uid=0(root)`, and both functions are part of the auto-discovered LLM tool registry, reachable from any anonymous TCP client to port 8899, allowing potential unauthorized access to sensitive files and system files.

Defensive priority

Defenders should prioritize patching the vulnerable Vibe-Trading AI tool to prevent unauthorized access to sensitive files. This involves updating to version 0.1.7 or later and verifying that the fixes address both the `safe_user_path()` and `read_document()` functions.

Recommended defensive actions

• Patch the Vibe-Trading AI tool to version 0.1.7 or later
• Verify that the fixes address both the `safe_user_path()` and `read_document()` functions
• Restrict access to the auto-discovered LLM tool registry
• Monitor for unauthorized access to sensitive files
• Review container configurations to prevent exposure
• Conduct regular security audits to identify potential vulnerabilities
• Implement additional logging and monitoring for suspicious activity

Evidence notes

The evidence provided indicates that the Vibe-Trading AI tool is vulnerable to file-read attacks due to the `safe_user_path()` and `read_document()` functions. The container runs as `uid=0(root)`, and the functions are part of the auto-discovered LLM tool registry, reachable from any anonymous TCP client to port 8899.

Sources and references

Methodology and review provenance

AI-assisted synthesis based on stored public vulnerability evidence. System validation, approval state, and publication status do not by themselves establish human review of this revision. PatchSiren helps prioritize defensive review and does not prove exposure or remediation on any system.