CVE-2026-71463: Jinja Template Information Disclosure

CVE-2026-71463 Overview

CVE-2026-71463 is an information disclosure vulnerability in a notification template subsystem that uses a Jinja Abstract Syntax Tree (AST) whitelist to gate template rendering. The whitelist only inspects static Getattr nodes, allowing dynamic subscript expressions such as job['job'+'_env'] and conditional gating like {% if job.id > 100 %} to bypass both the AST check and the pre-render stub validation. At runtime, exceptions from the gated branch are written as full Python tracebacks into the notification body and POSTed to an attacker-controlled webhook URL. The leaked data includes install paths, Python version, and source line numbers ([CWE-209]).

Critical Impact
Authenticated attackers with template-authoring privileges can exfiltrate runtime environment details to arbitrary webhook endpoints through crafted notification templates.

Affected Products

Red Hat products referenced in RHSA-2026:71113 and RHSA-2026:71114
Red Hat products referenced in RHSA-2026:71177 and RHSA-2026:71179
Notification template component using the Jinja AST whitelist validator

Discovery Timeline

2026-09-23 - CVE-2026-71463 published to NVD
2026-09-24 - Last updated in NVD database

Technical Details for CVE-2026-71463

Vulnerability Analysis

The notification subsystem attempts to sandbox user-supplied Jinja templates by walking the compiled AST and enforcing a whitelist of allowed attribute accesses. The walker only inspects Getattr nodes, which represent static dotted access such as job.name. Dynamic subscript access using Getitem nodes and string concatenation, for example job['job'+'_env'], is not evaluated against the whitelist.

A secondary defense renders the template against a stub context to catch runtime errors before deployment. The stub populates job.id with a small value, so branches gated by {% if job.id > 100 %} are never executed during validation. Once the template runs against real job data, the gated branch reaches sensitive attributes and can raise exceptions.

The notification renderer captures these exceptions and embeds full Python tracebacks into the notification body. The body is then POSTed to the webhook URL configured on the notification target, which the attacker controls.

Root Cause

The root cause is incomplete AST validation combined with error handling that exposes sensitive information [CWE-209]. The whitelist logic assumes attribute access is always expressed as a static Getattr, ignoring semantically equivalent Getitem and computed key expressions. The stub-based render check compounds the flaw by using deterministic values that never trigger conditional branches.

Attack Vector

An authenticated user with permission to author notification templates and configure webhook targets crafts a template that bypasses the whitelist through dynamic subscripts or conditional gating. When the notification fires against production job data, the runtime exception is serialized into the outbound HTTP POST body, delivering install paths, Python interpreter version, and source line numbers to the attacker's endpoint.

See the Red Hat CVE Analysis CVE-2026-71463 and Red Hat Bug Report #2512372 for further technical context.

Detection Methods for CVE-2026-71463

Indicators of Compromise

Outbound HTTP POST requests from the notification service to unfamiliar or newly created webhook domains.
Notification bodies containing Python traceback markers such as Traceback (most recent call last), File "/, or line followed by numeric identifiers.
Notification templates containing dynamic subscript patterns such as job[ combined with string concatenation, or conditional blocks referencing numeric ID thresholds.

Detection Strategies

Inspect stored notification templates for Getitem expressions and {% if %} blocks that reference identity or ID fields, and flag templates whose validation history differs from their runtime behavior.
Log and review all webhook target URLs configured by non-administrative users, correlating template edits with target creation events.
Alert on notification payloads whose size or content structure deviates from established baselines for the same template.

Monitoring Recommendations

Forward notification service egress logs and template modification events to a centralized analytics platform for correlation with user activity.
Track first-seen webhook destinations per tenant and generate alerts when new destinations receive traceback-shaped payloads.
Monitor for repeated notification failures on a single template, which may indicate an attacker iterating on payload structure.

How to Mitigate CVE-2026-71463

Immediate Actions Required

Apply vendor updates referenced in RHSA-2026:71113, RHSA-2026:71114, RHSA-2026:71177, and RHSA-2026:71179.
Audit existing notification templates for dynamic subscripts and conditional gating, and remove or rewrite any templates authored by non-trusted users.
Review all configured webhook destinations and revoke targets pointing to external or unmanaged infrastructure.

Patch Information

Red Hat has issued security advisories addressing CVE-2026-71463. See RHSA-2026:71113, RHSA-2026:71114, RHSA-2026:71177, and RHSA-2026:71179 for package versions and installation guidance. The fixes tighten AST validation to cover Getitem nodes and remove tracebacks from outbound notification bodies.

Workarounds

Restrict notification template authoring to administrative accounts until patches are applied.
Disable outbound webhook targets or restrict them to an allow-list of internal endpoints.
Configure the notification service to replace exception details with a generic error message before dispatch.

bash

# Configuration example: restrict webhook egress to an internal allow-list
# (adapt to your environment's firewall or egress proxy)
iptables -A OUTPUT -m owner --uid-owner notifier -d 10.0.0.0/8 -p tcp --dport 443 -j ACCEPT
iptables -A OUTPUT -m owner --uid-owner notifier -p tcp --dport 443 -j REJECT