CVE-2026-54654

Improper neutralization of carriage returns in comments allows code injection.

• CVSS 7.8
• 94
• Input Validation and Sanitization
• Local

datamodel-code-generator generates Python data models from schema definitions. From 0.14.1 until 0.60.2, the –extra-template-data comment field is rendered into Python comments in src/datamodel_code_generator/model/template/TypeAliasAnnotation.jinja2, src/datamodel_code_generator/model/template/TypedDict.jinja2, src/datamodel_code_generator/model/template/dataclass.jinja2, src/datamodel_code_generator/model/template/msgspec.Struct.jinja2, src/datamodel_code_generator/model/template/pydantic/BaseModel.jinja2, and src/datamodel_code_generator/model/template/pydantic_v2/BaseModel.jinja2 without neutralizing carriage returns in Python # comments, allowing an attacker-controlled comment value to inject Python code into generated models that runs when imported. This issue is fixed in version 0.60.2.

CVE

CVE-2026-54654

CWE

94

CVSS base score

7.8

Published

2026-07-28

OWASP

A03 Injection

Orthogonal defect classification

Checking

Code defect classification

Missing Check

Category

Input Validation and Sanitization

Subcategory

Command Injection

Accessibility scope

Local

Impact

Arbitrary Code Execution

Affected component

datamodel-co

Fixed by upgrading

Yes

Solution

Upgrade `datamodel-code-generator` to version 0.60.2 or later.

import jinja2

def generate_model(extra_data: dict) -> str:
    """
    Simulates vulnerable code generation from a Jinja2 template,
    analogous to datamodel-code-generator before version 0.60.2.
    """
    env = jinja2.Environment(trim_blocks=True, lstrip_blocks=True)
    template_str = """
{%- if extra_template_data.comment -%}
# {{ extra_template_data.comment }}
{%- endif -%}
class GeneratedModel:
    pass
"""
    template = env.from_string(template_str)

    # VULNERABLE: The comment is rendered directly, allowing newline characters to break out of the comment and inject code.
    return template.render(extra_template_data=extra_data)

import jinja2
from textwrap import indent

def generate_model(extra_data: dict) -> str:
    """
    Simulates fixed code generation from a Jinja2 template,
    analogous to datamodel-code-generator version 0.60.2 and later.
    """
    env = jinja2.Environment(trim_blocks=True, lstrip_blocks=True)
    template_str = """
{%- if extra_template_data.comment -%}
{{ extra_template_data.comment | comment_out }}
{%- endif -%}
class GeneratedModel:
    pass
"""
    template = env.from_string(template_str)

    # FIX: A filter is used in the template to sanitize the comment, prefixing each line with '# '.
    template.environment.filters['comment_out'] = lambda s: indent(s, '# ').strip()
    return template.render(extra_template_data=extra_data)

owned\nimport os;os.system('id')

@misc{vaitp:cve202654654,
  title        = {{Improper neutralization of carriage returns in comments allows code injection.}},
  author       = {Bogaerts, Fr\'ed\'eric and Ivaki, Naghmeh and Fonseca, Jos\'e},
  year         = {2026},
  note         = {VAITP Python Vulnerability Dataset, entry CVE-2026-54654},
  howpublished = {\url{https://netpack.pt/vaitp/vulnerability/CVE-2026-54654/}}
}