Skip to main content

LoLLMs CVE-2026-10595

| EUVDEUVD-2026-54864 HIGH
Relative Path Traversal (CWE-23)
2026-08-09 @huntr_ai GHSA-37c2-pxv8-c2mx
7.5
CVSS 3.0 · Vendor: huntr_ai
View

Attack ChainAIDerived

Hypothetical attack flow derived from CVE metadata – not a confirmed exploit chain

  1. Access
    Send crafted HTTP GET with %2e%2e path
  2. Delivery
    Bypass Starlette percent-encoding normalization
  3. Exploit
    pathlib.resolve() decodes traversal sequence
  4. Execution
    Escape STATIC_DIR filesystem boundary
  5. Persist
    Server returns arbitrary file contents
  6. Impact
    Exfiltrate credentials or configuration secrets

Vulnerability Assessment

Weakness
CWE
OWASP Top 10:2025
A01:2025 - Broken Access Control
ASVS 5.0
V5 - File Handling Relative path traversal prevention
Exploitation Conditions

No special conditions are required - the CVSS vector PR:N/UI:N/AC:L confirms unauthenticated, zero-interaction exploitation against any network-reachable LoLLMs 2.1.0 instance running the default configuration. The vulnerability is triggered via standard HTTP GET requests to the SPA catch-all route, which is active in the default deployment. The primary limiting factor is filesystem privilege: an attacker can only read files accessible to the OS account running the LoLLMs process, so sensitive material outside that account's read permissions is not exposed. The vulnerability does not provide write access, code execution, or privilege escalation through this vector alone. Deployments that are air-gapped, bound to localhost only, or placed behind authentication-enforcing reverse proxies have a substantially reduced attack surface despite the flaw being present in the code.

Impact

The CVSS 3.0 vector AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:N/A:N accurately reflects the exploitability profile: the attack is fully remote, requires zero authentication or user interaction, and has low complexity - all characteristics that make it trivially automatable. Confidentiality impact is rated High, meaning arbitrary file disclosure up to the privilege level of the LoLLMs process account. No EPSS data was present in the provided intelligence, so exploitation probability cannot be quantified. The vulnerability is not listed in the CISA KEV catalog, indicating no confirmed active exploitation at time of analysis. However, the public GitHub commit documents the exact vulnerable code and the bypass technique (%2e%2e), which functions as de facto exploit documentation. The Scope:Unchanged rating (S:U) correctly limits the blast radius - an attacker can read files the process user can read, but cannot directly pivot to other systems through this flaw alone. Real-world severity depends heavily on what sensitive material is readable by the LoLLMs process: configuration files containing API keys, credentials, or private keys would represent a critical secondary impact beyond the raw CVSS score suggests.

Exploit ScenarioDerived

An unauthenticated attacker sends an HTTP GET request to the LoLLMs web interface with a path containing URL-encoded traversal sequences, for example GET /%2e%2e/%2e%2e/%2e%2e/etc/passwd HTTP/1.1. Starlette's routing passes the undecoded path to the SPA catch-all handler, where pathlib resolves it to /etc/passwd outside the static directory, and the file contents are returned directly in the HTTP response. The public fix commit documents both the vulnerable code and the bypass technique, providing an effective blueprint for exploit construction.

This Week Schedule remediation this week. Why: HIGH severity (CVSS 7.0-8.9).
Patch available 0.5% exploitation probability (EPSS)

Get an e-mail when this CVE is exploited, patched or re-scored. Free account.

Alert me

AnalysisAIv1 – Aug 09, 2026

Path traversal in parisneo/lollms 2.1.0 exposes arbitrary server files to unauthenticated remote attackers via the SPA catch-all route in backend/routers/ui.py. URL-encoded dot-dot sequences (%2e%2e) bypass Starlette's built-in path normalization and are decoded by Python's pathlib.resolve(), allowing directory escape from the static file root with no privileges or user interaction required. No public exploit has been independently confirmed at time of analysis, though the public fix commit at GitHub fully documents the vulnerable code pattern and the bypass technique, substantially lowering the bar for exploit development.

RemediationAI

Upgrade to LoLLMs version 3 or later, which enforces filesystem containment via Path.resolve() and is_relative_to() in the SPA catch-all route. The patch is available at https://github.com/parisneo/lollms/commit/9bc6431ae7b708da76d51e7626a7cf48ff2b1d24 and full disclosure at https://huntr.com/bounties/af3f40a3-38a1-49d8-a5ae-3d72b9b05b98. If an immediate upgrade is not feasible, place a reverse proxy such as nginx in front of the LoLLMs instance and configure request filtering to reject or normalize URLs containing percent-encoded dot sequences (%2e) before they reach the Python application - note this approach adds operational complexity and may not catch all multi-encoding variants, making it an imperfect compensating control. A more reliable interim measure is restricting network access to the LoLLMs web interface to trusted hosts only (firewall or network segmentation), which eliminates unauthenticated remote exposure while upgrade is coordinated. Running the LoLLMs process under a dedicated low-privilege OS account with minimal filesystem read permissions limits the blast radius of exploitation even if the flaw is triggered.

Affected VersionsENISA

parisneo/lollms unspecified <3

Affected Packages

euvd parisneo_lollms_unspecified fix: 3

Source: OSV (Open Source Vulnerabilities)

Affects your stack

Affects 1 package in euvd

Checking this CVE against your own stacks and servers is part of the Basic plan and up.

Technical ContextAI · show details

LoLLMs is a Python-based AI assistant and LLM orchestration platform built on FastAPI/Starlette. The affected component is the SPA catch-all route, which serves static frontend assets by joining a user-controlled path segment directly onto STATIC_DIR using pathlib's / operator. The flaw emerges from a mismatch between two layers: Starlette's routing normalizes standard ../ sequences but does not decode percent-encoded equivalents such as %2e%2e/ before passing them to application logic. Python's pathlib.Path.resolve() does perform percent-decoding, so STATIC_DIR / '%2e%2e/etc/passwd' resolves to an absolute path outside the intended directory root. CWE-23 (Relative Path Traversal) precisely describes this root cause class - unsanitized user-controlled relative path components used in filesystem operations without containment enforcement. The fix, visible in commit 9bc6431, introduces Path.resolve() on both the candidate path and the base directory followed by an is_relative_to() check to enforce containment before serving any file.

Know when Python gets its next vulnerability

  • Watch products and get an email when something new is published
  • Check this CVE against your own stack and servers
  • Exploit and PoC references, threat intelligence from 22 sources

No credit card · 7-day full trial

Severity by source

Vendor (huntr_ai) PRIMARY 7.5 HIGH
AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:N/A:N
vuln.today AI 7.5 HIGH

Network-reachable HTTP endpoint (AV:N), no authentication or configuration prerequisite (PR:N, AC:L), read-only arbitrary file disclosure (C:H, I:N, A:N).

3.1 AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:N/A:N
4.0 AV:N/AC:L/AT:N/PR:N/UI:N/VC:H/VI:N/VA:N/SC:N/SI:N/SA:N

Primary rating from Vendor (huntr_ai).

CVSS VectorVendor: huntr_ai

Attack Vector
Network
Attack Complexity
Low
Privileges Required
None
User Interaction
None
Scope
Unchanged
Confidentiality
High
Integrity
None
Availability
None

Lifecycle Timeline

2
Patch available
Aug 09, 2026 - 05:01 EUVD
CVE Published
Aug 09, 2026 - 03:52 cve.org
HIGH 7.5

DescriptionCVE.org

A path traversal vulnerability exists in parisneo/lollms version 2.1.0, specifically in the SPA catch-all route implemented in backend/routers/ui.py. The vulnerability arises from the improper handling of user-controlled path input, which is directly joined into a filesystem path without sanitization or containment checks. URL-encoded dot-dot sequences (%2e%2e) bypass Starlette's built-in path normalization and are resolved by Python's pathlib, allowing an unauthenticated attacker to read arbitrary files on the server. This issue has been resolved in version 3.

Executive ViewAI

Why It Matters

ParisNeo's lollms 2.1.0 contains a path traversal vulnerability enabling unauthenticated remote attackers to read arbitrary files from affected servers, creating direct exposure to sensitive credentials, configurations, and proprietary data. Organizations operating this version face immediate confidentiality breach risk and require emergency containment before potential lateral movement or data exfiltration occurs.

Business Impact: High
Compensating Controls

No vendor-released patch identified at time of analysis. Restrict network access to lollms 2.1.0 services through firewall rules, limiting exposure to trusted internal networks only and disabling internet-facing access where operationally feasible. Deploy Web Application Firewall rules to block path traversal patterns. Increase access logging and monitoring for exploitation attempts using encoded dot-dot sequences. Maintain asset inventory of affected instances and prepare parallel migration planning to alternative solutions while monitoring vendor patch progress.

Threat Intelligence Overview

EPSS Score
0.5%
Top 60% exploitable
Prevalence
Low
niche AI assistant platform, limited enterprise production footprint
Social Media
1
Sightings
Data Quality
6 sources
Confidence: High

How this decision was reached

The engine reads each signal below and applies a fixed list of rules from top to bottom. The first rule that matches sets the recommended action – like a statement, every line is shown.

SignalValue
Severity HIGH
CISA KEV No
Public POC No
EPSS 0.5%
Patch available Yes
Active threat No
Deciding rule
HIGH severity (CVSS 7.0-8.9)
What every action level means
Emergency Immediate action required. Criteria: CRITICAL + in CISA KEV (actively exploited) without patch, OR CRITICAL + EPSS > 30% without patch, OR CRITICAL + active threat signals without patch.
Act Now Prioritize within 24-48h. Criteria: in KEV with patch available, OR CRITICAL + public POC, OR HIGH + EPSS > 10%, OR CRITICAL + active threat, OR any CRITICAL severity.
This Week Schedule remediation this week. Criteria: HIGH severity (CVSS 7.0-8.9). Patch within normal change window.
This Month Plan within 30 days. Criteria: MEDIUM severity (CVSS 4.0-6.9), lower exploitation probability. Include in regular patch cycle.
Monitor Track and reassess. Criteria: LOW severity (CVSS 0.1-3.9), minimal exploitation risk. No immediate action needed.

SSVC AssessmentENISA

SSVC Decision· derived
Scheduled
Exploitation· derived
None
System Exposure· derived
Open
Automatable· derived
Yes
Human Impact
High

Priority Score

38
Low Medium High Critical
KEV: 0
EPSS: 0
CVSS: +38
PoC: 0
Weaponized: 0
Threat intel: 0
Gov advisory: 0

Attack Technique

Path traversal exploits occur when applications use user-controlled input to construct file system paths without proper validation.

More in Python

View all
CVE-2025-24016 CRITICAL POC
9.9 Feb 10

Wazuh SIEM platform versions 4.4.0 through 4.9.0 contain an unsafe deserialization vulnerability in the DistributedAPI t

CVE-2025-27520 CRITICAL POC
9.8 Apr 04

BentoML version 1.4.2 and earlier contains an unauthenticated remote code execution vulnerability through insecure deser

CVE-2025-2945 CRITICAL POC
9.9 Apr 03

pgAdmin 4 contains critical remote code execution vulnerabilities in the Query Tool download and Cloud Deployment endpoi

CVE-2013-5093 MEDIUM POC
6.8 Sep 27

The renderLocalView function in render/views.py in graphite-web in Graphite 0.9.5 through 0.9.10 uses the pickle Python

CVE-2025-32375 CRITICAL POC
9.8 Apr 09

BentoML is a Python library for building online serving systems optimized for AI apps and model inference. Rated critica

CVE-2014-0224 HIGH POC
7.4 Jun 05

OpenSSL before 0.9.8za, 1.0.0 before 1.0.0m, and 1.0.1 before 1.0.1h does not properly restrict processing of ChangeCiph

CVE-2024-21644 HIGH POC
7.5 Jan 08

pyLoad download manager version prior to 0.5.0b3.dev77 exposes the Flask SECRET_KEY through an unauthenticated endpoint.

CVE-2026-33017 CRITICAL POC
9.3 Mar 17

Langflow (a visual LLM pipeline builder) contains a critical unauthenticated code execution vulnerability (CVE-2026-3301

CVE-2017-9462 HIGH POC
8.8 Jun 06

In Mercurial before 4.1.3, "hg serve --stdio" allows remote authenticated users to launch the Python debugger, and conse

CVE-2026-49869 CRITICAL POC
10.0 Jun 26

Unauthenticated remote code execution affects Kestra OSS (the open-source event-driven orchestration platform) prior to

CVE-2026-39987 CRITICAL POC
9.3 Apr 08

Unauthenticated remote code execution in Marimo ≤0.20.4 allows attackers to execute arbitrary system commands via the `/

CVE-2024-21645 MEDIUM POC
5.3 Jan 08

pyLoad is the free and open-source Download Manager written in pure Python. Rated medium severity (CVSS 5.3), this vulne

This site uses cookies essential for authentication and security. No tracking or analytics cookies are used. Privacy Policy