CVE-2026-33149 is a host header poisoning vulnerability in Tandoor Recipes affecting versions up to and including 2.5.3. The application is configured with a permissive default host validation setting that accepts arbitrary HTTP Host header values. Because the application uses Django's absolute-URL generation functionality to construct server-generated links in multiple contexts, including invitation emails, API pagination, and OpenAPI schema generation, an attacker can supply a crafted Host header and cause the application to embed attacker-controlled origins in generated URLs. The most severe documented case is invite link poisoning: when an administrator creates an invite, the resulting email can contain a link pointing to attacker-controlled infrastructure rather than the legitimate application, allowing interception of the invite token when the recipient follows the link.
Mallory correlates every CVE against your assets, your vendors, and active adversary campaigns. Know which vulnerabilities matter for you, not just which ones are loud.
What it means. What to do now. Patch path, mitigations, and the assume-compromise checklist.
What an attacker gets, and what they’ve been doing with it.
If you can’t patch tonight, do this now.
Patch, then assume compromise.
1 valid exploit after Mallory filtered fakes, detection scripts, and README-only repos.
This repository is a small standalone Python proof-of-concept for CVE-2026-33149 affecting Tandoor Recipes <= 2.5.3. It contains three files: an MIT license, a detailed README describing the vulnerability and attack scenarios, and a single executable script, poc.py, which is the main exploit entry point. The exploit targets a Host header injection condition caused by permissive Django host validation (ALLOWED_HOSTS='*'). Its core capability is to send authenticated requests with an attacker-controlled Host header and observe whether Tandoor Recipes uses that untrusted host when generating absolute URLs. The script is modular and supports multiple checks: confirming arbitrary Host acceptance, testing whether API pagination links reflect the injected host, testing whether schema/OpenAPI output reflects the injected host, and triggering invite poisoning so that invite emails contain attacker-controlled links. Operationally, the script uses the requests library, disables TLS verification, and supports two authentication modes: session cookie + CSRF token, or HTTP Basic Auth. The attacker supplies a target base URL and an attacker-controlled domain via command-line arguments. Optional arguments include victim email and group ID for the invite-poisoning workflow. The script prints colored status output and a final vulnerability summary. This is not merely a detector: it actively exercises vulnerable functionality and can trigger a real security impact by causing poisoned invite links to be generated. However, it does not deliver a post-exploitation shell or arbitrary code execution payload; instead, its payload is the malicious Host header value used to redirect server-generated URLs. Overall, the repository is a focused, operational web exploit PoC demonstrating URL poisoning and phishing/token-capture implications in Tandoor Recipes.
Products and vendors Mallory has correlated with this vulnerability. Open in Mallory to drill down to specific CPE configurations and version ranges.
Vendor-confirmed product mapping. Mallory continuously reconciles this list against your asset inventory.
5 sources tracked across advisories and community write-ups. News coverage will land here when it surfaces.
No news coverage yet. Advisories and community discussion only.
Query your assets running an affected version, and investigate the blast radius.
Every observed campaign linking this CVE to a named adversary.
Malware families riding this exploit, with evidence and IOCs.
YARA, Sigma, Snort, and vendor rules, auto-deployed to your SIEM.
Cross-references every affected SKU, including bundled OEM variants.
Community discussion across Reddit, Mastodon, and other social sources.