# CVE-2026-120313: Network-AI path traversal in InputSanitizer.sanitizePath

Source: https://junglewise.ai/threats/cve-2026-120313-network-ai-before-5-15-5-contains-a-path-traversal-vulnerability  
Publisher: Junglewise Threat Intelligence, https://junglewise.ai/threats  
Published: 2026-11-02. Updated: 2026-11-02  
License: CC BY 4.0 with a link back

CVE-2026-120313 · Severity: high · CVSS 7.5 · Published 2026-11-02

Network-AI is a library that applications use to validate file paths for secure operations within a restricted directory sandbox. A flaw in its path validation function allows attackers to read, list, or write files outside the intended sandbox by supplying paths to sibling directories with names that share a prefix with the base directory name. This can expose sensitive files or allow unauthorized file creation.

Technologies: [Jovancoding Network-AI](https://junglewise.ai/threats/technologies/network-ai). Vendors: [Jovancoding](https://junglewise.ai/threats/vendors/jovancoding).

## Quick answers

**Is CVE-2026-120313 exploited in the wild?** No. Junglewise Threat Intelligence has no report of CVE-2026-120313 being exploited in the wild.

**How severe is CVE-2026-120313?** High severity, with a CVSS score of 7.5, published 2 November 2026.

**Which products and versions are affected?** Jovancoding Network-AI before 5.15.5.

**Is there a fix for CVE-2026-120313?** Yes: fixed in 5.15.5. Junglewise finds every place you run it, drafts the upgrade and applies it once you approve.

**How is CVE-2026-120313 attacked?** It is attacked over the network, in a software library.

**Am I affected by CVE-2026-120313?** Check whether you run Jovancoding Network-AI 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.

## How to fix

- Fixed in 5.15.5

## Technical details

InputSanitizer.sanitizePath() uses a bare string prefix check (startsWith) without path-separator anchoring, allowing sibling paths like /data/project-evil to bypass validation against base /data/project. An attacker who controls the input path argument can supply absolute paths or relative traversals that resolve to out-of-scope siblings and are returned as validated, enabling read, write, and directory-listing operations outside the sandbox boundary.

## Affected products

- Jovancoding Network-AI before 5.15.5

## Timeline

- 2026-10-17: disclosed
- 2026-11-02: advisory

## References

- [https://github.com/Jovancoding/Network-AI/security/advisories/GHSA-5mm5-4mh4-6gm6](https://github.com/Jovancoding/Network-AI/security/advisories/GHSA-5mm5-4mh4-6gm6)
- [https://www.vulncheck.com/advisories/network-ai-before-5.15.5-path-traversal-via-inputsanitizer-sanitizepath](https://www.vulncheck.com/advisories/network-ai-before-5.15.5-path-traversal-via-inputsanitizer-sanitizepath)

## Related threats

- [Network-AI A2AServer missing authentication in task operations](https://junglewise.ai/threats/cve-2026-120314-network-ai-from-5-14-0-before-5-15-5-contains-a-missing)
- [Network-AI AgentVCR path traversal in cassette name](https://junglewise.ai/threats/cve-2026-120312-network-ai-from-5-13-4-before-5-15-5-contains-a-path-traversal)
- [Network-AI FileJobStore path traversal in job ID handling](https://junglewise.ai/threats/cve-2026-120311-network-ai-from-5-13-4-before-5-15-5-contains-a-path-traversal)
- [Network-AI security control bypass in ClaudeHookBridge](https://junglewise.ai/threats/cve-2026-109331-network-ai-from-5-14-0-before-5-15-2-contains-a-security-control)
- [Network-AI DashboardServer cross-site WebSocket hijacking](https://junglewise.ai/threats/cve-2026-109330-network-ai-5-0-0-before-5-15-2-contains-a-cross-site-websocket)