
PoC for CVE-2026-59970 Blind SSRF in Tandoor Recipes, CWE-918
This repository contains a small proof-of-concept for authorized SSRF-style behavior associated with CVE-2026-59970, CVSS score 7.7 high, in the Tandoor Recipes application. The scripts are intended for security research, validation, and controlled testing in an environment you own or are explicitly authorized to test. At the the time of release this is the only available PoC for this CVE. This PoC is still work in progress and may not be fully functional, as it has not been tested yet. Only the ai-step-sort action works as inteded. This PoC will be updated later to be propely functional. This vulnerability is fixed in Recipes version 2.6.10
poc.py — Python script that authenticates to a Tandoor instance, obtains a CSRF token, and exercises the AI-related endpoints to observe how the application handles provider requests.callback.py — minimal HTTP listener that records inbound requests, including path, query string, and source IP, which is useful for confirming callback or SSRF behavior during validation.The goal of this PoC is to help confirm whether a Tandoor deployment is exposing outbound request behavior through an authenticated AI integration flow. It is intentionally simple and does not attempt to disguise itself or evade defensive controls.
Tandoor Recipes is an application for managing recipes, planning meals, and building shopping lists. Prior to 2.6.10, passes an administrator-controlled AI provider URL directly to LiteLLM without applying the server's outbound-request SSRF checks when an authenticated user creates a Space, configures an AI provider with a URL targeting an internal address, and invokes an AI action. In cookbook/views/api.py, the affected AI actions call completion through TARGET/api/ai-import/ and api/ai-step-sort/ while bypassing safe_request. The attack works because the server connects to the supplied provider URL from each AI action and exposes connection success, refusal, or timeout as a blind network-reachability oracle. This can reach internal services, other hosts on the server network, and cloud metadata endpoints and map reachable internal hosts and ports. CVE Feed
This project is for authorized security testing only. Do not run it against a system you do not own or have explicit permission to test.
requests Python packageInstall dependencies:
python3 -m pip install requests
python3 callback.py
The listener will bind to port 4444 by default and print details for incoming requests.
Add an ai provider URL in the application, this should be your callback URL for testing. Use this ID as the provider-id when running the PoC.
python3 poc.py \
--base-url "https://tandoor.example.com" \
--username "your-username" \
--provider-id 1 \
--action ai-step-sort
The PoC will:
login_path and any related fields accordingly.Use this code only for:
Do not use it for unauthorized access, production testing, or malicious activity.