CVE-2026-59925: Mistune Markdown Parser Denial of Service via Quadratic Parsing Complexity

Technical summary

The vulnerability resides in src/mistune/inline_parser.py, where the inline parser scans forward from every potential opening emphasis marker to find matching closing markers. On input containing long runs of well-formed emphasis pairs (e.g., **a****b****c**...), the parser exhibits O(n²) behavior because it re-scans overlapping candidate ranges. The quadratic complexity allows an attacker to craft payloads that cause parsing time to balloon from milliseconds to seconds or minutes. The issue is rooted in CWE-1333 (Inefficient Regular Expression Complexity) and CWE-407 (Algorithmic Complexity). No code execution or data breach risk exists; the attack surface is purely availability.

Business impact

Applications relying on Mistune to parse user-supplied or untrusted Markdown—such as documentation platforms, comment systems, wiki software, or static site generators—face resource exhaustion attacks. A single malicious Markdown file or POST request can spike CPU usage, degrading service for legitimate users or triggering auto-scaling costs in cloud environments. Widespread adoption of Mistune in Python projects means many organizations may be indirectly affected through dependencies.

Affected systems

Mistune versions prior to 3.3.0 are vulnerable. The vulnerability affects any application or service using Mistune for Markdown parsing, particularly those exposed to untrusted input (e.g., user comments, uploaded documents, API endpoints). This includes static site generators, documentation tools, and any Python project listing Mistune as a direct or transitive dependency.

Exploitability

Exploitability is high. The attack requires only network access (AV:N) and no authentication or user interaction (PR:N, UI:N). An attacker can trigger the vulnerability by sending a single HTTP request or API call containing the crafted Markdown payload. No special privileges, complex setup, or interaction with other vulnerabilities is needed. Public proof-of-concept construction is straightforward—a test can be performed locally to determine if a Mistune version is vulnerable.

Remediation

Upgrade Mistune to version 3.3.0 or later. This version includes algorithmic improvements to the inline parser that eliminate the quadratic behavior. For projects unable to upgrade immediately, input size limits or rate limiting on Markdown parsing requests can provide temporary mitigation, though they are not a substitute for patching.

Patch guidance

Apply version 3.3.0 or any subsequent release. Verify the patch by checking the Mistune version in your environment (e.g., `pip show mistune`) and confirming it reports 3.3.0 or higher. If Mistune is installed as a transitive dependency, update the direct dependency tree—most dependency managers (pip, poetry, pipenv) support `pip install --upgrade mistune` or equivalent. Test in a staging environment to ensure compatibility with your Markdown workflows before rolling out to production.

Detection guidance

Monitor for unusual CPU spikes or parser timeouts when processing Markdown input. In logs, look for parsing operations that take significantly longer than baseline (>5 seconds for small payloads). If feasible, implement input validation rules that reject Markdown with excessive consecutive emphasis markers (e.g., more than 10 consecutive asterisks). Vulnerability scanners can detect Mistune <3.3.0 in dependencies via Software Composition Analysis (SCA) tools. Track Mistune version in your Bill of Materials and flag instances below 3.3.0.

Why prioritize this

CVSS 7.5 (HIGH) reflects the ease of exploitation and moderate impact. While no confidentiality or integrity risk exists, the vulnerability is network-accessible, requires no authentication, and can reliably deny service to end-users or degrade availability for all Markdown parsing operations. Organizations parsing untrusted Markdown at scale should prioritize patching. Given the simplicity of the attack and the potential for widespread deployment, this warrants urgent review and patching cycles.

Risk score, explained

The CVSS 3.1 vector (AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H) scores HIGH (7.5) because: Attack Vector is Network (AV:N)—any remote attacker can trigger the flaw. Attack Complexity is Low (AC:L)—no special conditions or race conditions; a simple payload suffices. Privileges Required and User Interaction are None (PR:N, UI:N)—authentication and user clicks are unnecessary. Scope is Unchanged (S:U)—the impact is confined to the Mistune parser and its host process. Availability is High (A:H)—the parser can hang or consume all CPU, effectively denying service. Confidentiality and Integrity remain None (C:N, I:N).

Frequently asked questions