---
id: "CVE-2026-97701"
kind: "vulnerability"
aliases: ["CVE-2026-97701"]
title: "Zulip: unbounded Markdown render request lets one user slow the server"
severity: "medium"
cvss: 6.5
epss: null
exploited: false
status: "open"
category: "runtime"
products: []
ecosystems: []
tags: ["chat", "runtime", "self-hosted", "zulip"]
not_affected_if: ["already-on-12.2", "managed-hosting", "no-public-signup"]
published_at: "2026-11-04T19:17:41+00:00"
first_seen_at: "2026-11-04T19:28:49+00:00"
updated_at: "2026-11-04T19:30:15+00:00"
sources: [{"source": "nvd", "url": "https://nvd.nist.gov/vuln/detail/CVE-2026-97701", "first_seen_at": "2026-11-04T19:28:49+00:00"}]
---

# Zulip: unbounded Markdown render request lets one user slow the server

Zulip Server through version 12.1 has an authenticated endpoint, POST /messages/render, that renders Markdown without the message length limit used elsewhere. A logged-in user can send very large Markdown and, by repeating or running requests concurrently, use up backend CPU and degrade service for other users.

_Summary, check and fix written by a model from the sources; labelled as such._

## How to check

- Check the Zulip Server version, for example with `/etc/zulip/zulip.conf` or the admin page, and see whether it is `12.1` or earlier.
- Check whether your server is reachable by users who hold a regular account.

## Fix

- Zulip Server: move to `12.2`.

## Timeline

- 2026-11-04T19:28:49+00:00 nvd https://nvd.nist.gov/vuln/detail/CVE-2026-97701

## Documents

- [advisory] Vendor advisory: github.com https://github.com/zulip/zulip/security/advisories/GHSA-m257-w4r5-5rmf
- [fix] Patch: github.com https://github.com/zulip/zulip/commit/31773c77c501391ede0fa8ec727f39508664e0ae
- [fix] Patch: github.com https://github.com/zulip/zulip/commit/4d94dd2ee6337a4b3351694dd314a45d3fe11031
- [fix] Release notes: github.com https://github.com/zulip/zulip/releases/tag/12.2

## From the sources

Zulip is an open-source team collaboration tool. In Zulip Server versions through 12.1, the authenticated POST /messages/render endpoint renders the content parameter without enforcing the normal message length limit used by message send and edit operations. An authenticated user can submit Markdown up to the server body-size limit, causing disproportionate backend CPU consumption, and repeated or concurrent requests can degrade Markdown rendering performance or backend availability for other users. This issue is fixed in version 12.2.

---

_Free to read at a human pace. An API with keys and an MCP server are in preparation; early access: hello@infrastack.ai_