CVE-2025-6984: EverNoteLoader: XXE exposes host files in LangChain

Q: Is CVE-2025-6984 actively exploited?

Proof-of-concept exploit code is publicly available for CVE-2025-6984, increasing the risk of exploitation.

Q: How to fix CVE-2025-6984?

1. PATCH: Upgrade langchain-community to ≥0.3.27 immediately — the fix disables external entity resolution in etree.iterparse(). 2. AUDIT: Enumerate all uses of EverNoteLoader and any XML-parsing document loaders across your LangChain deployments. 3. INPUT VALIDATION: Reject .enex files from untrusted sources at API/upload boundaries before they reach the loader. 4. LEAST PRIVILEGE: Run LangChain processes with read-only mounts and no access to /etc, credential stores, ~/.aws, or .env files using container security contexts or seccomp profiles. 5. NETWORK EGRESS: Block outbound HTTP/file connections from application servers to prevent out-of-band XXE data exfiltration. 6. DETECTION: Alert on anomalous file read patterns in application logs (paths: /etc/passwd, /proc/self/environ, ~/.aws/credentials, *.env).

Q: What systems are affected by CVE-2025-6984?

This vulnerability affects the following AI/ML architecture patterns: RAG pipelines, document processing pipelines, agent frameworks, data ingestion pipelines.

Q: What is the CVSS score for CVE-2025-6984?

CVE-2025-6984 has a CVSS v3.1 base score of 7.5 (HIGH). The EPSS exploitation probability is 1.65%.

GHSA-pc6w-59fv-rh23HIGHPoC AVAILABLENUCLEI TEMPLATECISA: TRACK*

Published September 4, 2025

CISO Take

Any LangChain deployment ingesting Evernote files (.enex) from untrusted sources is vulnerable to local file disclosure via XXE — attackers can read /etc/passwd, .env files, and cloud credentials off the host. Patch langchain-community to 0.3.27 immediately. Until patched, treat all external Evernote file ingestion as a critical-risk operation and restrict filesystem access of the process running LangChain.

What is the risk?

High severity with near-zero exploitation complexity — XXE payloads are trivially crafted and require no authentication or user interaction (CVSS AV:N/AC:L/PR:N/UI:N). EPSS of 2% reflects limited active exploitation today, but LangChain's massive footprint in RAG and document processing pipelines makes this a high-value target with broad blast radius. Risk is elevated for organizations ingesting Evernote exports from third parties, customers, or automated workflows.

What systems are affected?

Package	Ecosystem	Vulnerable Range	Patched
LangChain Community	pip	< 0.3.27	`0.3.27`
148.8KOpenSSF 5.71.3K dependentsPushed 5d ago50% patched~48d to patchFull package profile →

Do you use LangChain Community? You're affected.

How severe is it?

CVSS 3.1

7.5 / 10

EPSS

1.6%

chance of exploitation in 30 days

Higher than 77% of all CVEs

Source: EPSS v3 — FIRST.org

Exploitation Status

Exploit Available

Exploitation: MEDIUM

Sophistication

Trivial

Exploitation Confidence

medium

○CISA SSVC: Public PoC

○Public PoC indexed (trickest/cve)

○Nuclei detection template available

Composite signal derived from CISA KEV, VulnCheck KEV, CISA SSVC, EPSS, Metasploit, Exploit-DB, trickest/cve, Nuclei templates, and inthewild.io exploitation reports.

What is the attack surface?

AVNetwork

ACLow

PRNone

UINone

SUnchanged

CHigh

INone

ANone

What should I do?

6 steps

PATCH
Upgrade langchain-community to ≥0.3.27 immediately — the fix disables external entity resolution in etree.iterparse().
AUDIT
Enumerate all uses of EverNoteLoader and any XML-parsing document loaders across your LangChain deployments.
INPUT VALIDATION
Reject .enex files from untrusted sources at API/upload boundaries before they reach the loader.
LEAST PRIVILEGE
Run LangChain processes with read-only mounts and no access to /etc, credential stores, ~/.aws, or .env files using container security contexts or seccomp profiles.
NETWORK EGRESS
Block outbound HTTP/file connections from application servers to prevent out-of-band XXE data exfiltration.
DETECTION
Alert on anomalous file read patterns in application logs (paths: /etc/passwd, /proc/self/environ, ~/.aws/credentials, *.env).

What does CISA's SSVC say?

DecisionTrack*

Exploitationpoc

AutomatableNo

Technical Impactpartial

Source: CISA Vulnrichment (SSVC v2.0). Decision based on the CISA Coordinator decision tree.

How is it classified?

Data ExtractionData LeakageFrameworkRAGAML.T0010.001- AI Software AML.T0025- Exfiltration via Cyber Means AML.T0037- Data from Local System AML.T0049- Exploit Public-Facing Application

Which compliance frameworks are affected?

This CVE is relevant to:

EU AI Act

Article 9-Risk management system

ISO 42001

A.6.2.6-AI system security

NIST AI RMF

MANAGE 2.2-Treatments, responses, and chosen actions are documented and monitored regularly

OWASP LLM Top 10

LLM05-Supply Chain VulnerabilitiesLLM06-Sensitive Information Disclosure

Frequently Asked Questions

What is CVE-2025-6984?

Is CVE-2025-6984 actively exploited?

Proof-of-concept exploit code is publicly available for CVE-2025-6984, increasing the risk of exploitation.

How to fix CVE-2025-6984?

1. PATCH: Upgrade langchain-community to ≥0.3.27 immediately — the fix disables external entity resolution in etree.iterparse(). 2. AUDIT: Enumerate all uses of EverNoteLoader and any XML-parsing document loaders across your LangChain deployments. 3. INPUT VALIDATION: Reject .enex files from untrusted sources at API/upload boundaries before they reach the loader. 4. LEAST PRIVILEGE: Run LangChain processes with read-only mounts and no access to /etc, credential stores, ~/.aws, or .env files using container security contexts or seccomp profiles. 5. NETWORK EGRESS: Block outbound HTTP/file connections from application servers to prevent out-of-band XXE data exfiltration. 6. DETECTION: Alert on anomalous file read patterns in application logs (paths: /etc/passwd, /proc/self/environ, ~/.aws/credentials, *.env).

What systems are affected by CVE-2025-6984?

This vulnerability affects the following AI/ML architecture patterns: RAG pipelines, document processing pipelines, agent frameworks, data ingestion pipelines.

What is the CVSS score for CVE-2025-6984?

CVE-2025-6984 has a CVSS v3.1 base score of 7.5 (HIGH). The EPSS exploitation probability is 1.65%.

What is the AI security impact?

Affected AI Architectures

RAG pipelinesdocument processing pipelinesagent frameworksdata ingestion pipelines

MITRE ATLAS Techniques

AML.T0010.001AI Software

AML.T0025Exfiltration via Cyber Means

AML.T0037Data from Local System

AML.T0049Exploit Public-Facing Application

Compliance Controls Affected

EU AI Act: Article 9

ISO 42001: A.6.2.6

NIST AI RMF: MANAGE 2.2

OWASP LLM Top 10: LLM05, LLM06

What are the technical details?

Original Advisory

The langchain-ai/langchain project, specifically the EverNoteLoader component, is vulnerable to XML External Entity (XXE) attacks due to insecure XML parsing. The affected version is 0.3.63. The vulnerability arises from the use of etree.iterparse() without disabling external entity references, which can lead to sensitive information disclosure. An attacker could exploit this by crafting a malicious XML payload that references local files, potentially exposing sensitive data such as /etc/passwd.

Exploitation Scenario

An attacker uploads a crafted Evernote export file to a RAG-based document ingestion endpoint. The .enex file embeds an XXE payload: <!DOCTYPE foo [<!ENTITY xxe SYSTEM 'file:///etc/passwd'>]> referencing a local file. When EverNoteLoader processes it via unpatched etree.iterparse(), the external entity resolves and injects /etc/passwd contents into the parsed document. The attacker reads the exfiltrated data through the application response or infers it via error messages. They then iterate through high-value paths (/proc/self/environ, ~/.aws/credentials, /run/secrets/) to harvest cloud credentials. On a Kubernetes pod or cloud VM, this yields IAM tokens sufficient for lateral movement into the broader cloud environment.

Weaknesses (CWE)

CWE-200Exposure of Sensitive Information to an Unauthorized ActorPrimary CWE-611Improper Restriction of XML External Entity ReferencePrimary CWE-200Exposure of Sensitive Information to an Unauthorized Actor

CWE-200 — Exposure of Sensitive Information to an Unauthorized Actor: The product exposes sensitive information to an actor that is not explicitly authorized to have access to that information.

[Architecture and Design] Compartmentalize the system to have "safe" areas where trust boundaries can be unambiguously drawn. Do not allow sensitive data to go outside of the trust boundary and always be careful when interfacing with a compartment outside of the safe area. Ensure that appropriate compartmentalization is built into the system design, and the compartmentalization allows for and reinforces privilege separation functionality. Architects and designers should rely on the principle of least privilege to decide the appropriate time to use privileges and the time to drop privileges.

Source: MITRE CWE corpus.