# CVE-2026-77294: TREK server-side request forgery in LLM settings

Source: https://junglewise.ai/threats/cve-2026-77294-trek-is-a-collaborative-travel-planner-prior-to-3-3-0-trek-allows  
Publisher: Junglewise Threat Intelligence, https://junglewise.ai/threats  
Published: 2026-09-24. Updated: 2026-09-26  
License: CC BY 4.0 with a link back

CVE-2026-77294 · Severity: high · CVSS 8.1 · Published 2026-09-24

TREK is a collaborative travel planner that supports AI-assisted trip planning. Prior to version 3.3.0, an authenticated user can inject a malicious LLM server URL through the settings API, causing the server to make requests to attacker-controlled or internal services. This allows attackers to discover internal infrastructure, access cloud metadata endpoints, and potentially steal credentials to modify protected cloud resources.

Technologies: [Liketrek TREK](https://junglewise.ai/threats/technologies/liketrek-trek). Vendors: [Liketrek](https://junglewise.ai/threats/vendors/liketrek).

## Quick answers

**Is CVE-2026-77294 exploited in the wild?** No. Junglewise Threat Intelligence has no report of CVE-2026-77294 being exploited in the wild.

**How severe is CVE-2026-77294?** High severity, with a CVSS score of 8.1, published 24 September 2026.

**Which products and versions are affected?** Liketrek TREK before 3.3.0.

**Is there a fix for CVE-2026-77294?** A patch was published on 2026-07-11: Fixed in version 3.3.0 Junglewise finds every place you run it, drafts the upgrade and applies it once you approve.

**How is CVE-2026-77294 attacked?** It is attacked over the network, in a web application.

**Am I affected by CVE-2026-77294?** Check whether you run Liketrek TREK 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

An authenticated user with write permission to a trip instance can store an attacker-controlled llm_base_url through the settings API when the LLM_PARSING feature is enabled. The clients in the OpenAI-compatible, Anthropic, and Ollama format modules consume this value without server-side request forgery (SSRF) guards. When AI-assisted trip parsing is triggered, the server requests the supplied destination and returns upstream error responses in parsing warnings, disclosing internal service information and cloud metadata.

## Affected products

- liketrek TREK before 3.3.0

## Timeline

- 2026-09-24: disclosed
- 2026-07-11: patched: Fixed in version 3.3.0

## References

- [https://github.com/liketrek/TREK/commit/19064b39176660f3be3a2df198c87949504046e2](https://github.com/liketrek/TREK/commit/19064b39176660f3be3a2df198c87949504046e2)
- [https://github.com/liketrek/TREK/pull/1520](https://github.com/liketrek/TREK/pull/1520)

## Related threats

- [TREK SSRF bypass via IPv6 transition addresses](https://junglewise.ai/threats/cve-2026-85738-trek-is-a-collaborative-travel-planner-prior-to-3-4-0-the)
- [TREK get_trip_summary missing scope validation](https://junglewise.ai/threats/cve-2026-77321-trek-is-a-collaborative-travel-planner-prior-to-3-3-0-the-get)
- [TREK collaborative travel planner information disclosure in shared trip API](https://junglewise.ai/threats/cve-2026-77320-trek-is-a-collaborative-travel-planner-prior-to-3-3-0)