Sign in
Deutsch English Español Polski
High

RunGptLLM class in LlamaIndex has a command injection

  • Disclosed: 16.05.2024
  • Updated: 11.10.2026 13:19

A command injection vulnerability exists in the RunGptLLM class of the llama_index library, version 0.9.47, used by the RunGpt framework from JinaAI to connect to Language Learning Models (LLMs). The vulnerability arises from the improper use of the eval function, allowing a malicious or compromised LLM hosting provider to execute arbitrary commands on the client's machine. This issue was fixed in version 0.10.13. The exploitation of this vulnerability could lead to a hosting provider gaining full control over client machines.

Affected software

Click a technology to see all vulnerabilities affecting it.

PlatformPackageVulnerable versionsFixed in
Python (PyPI) llama-index >= 0.9.47, < 0.10.13 0.10.13
Python (PyPI) llama-index-llms-rungpt < 0.1.3 0.1.3

References

CVSS score 8.8/10
CVSS vector
CVSS:3.0/AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H
Weakness (CWE)
CWE-94
Platform
Python (PyPI)
Package
llama-index, llama-index-llms-rungpt
Disclosed
16.05.2024
Updated
11.10.2026 13:19

Sources and licenses

OSV / GitHub Advisory Database - PYSEC-2024-324 - PyPA Advisory Database; CC BY 4.0

OSV / GitHub Advisory Database - GHSA-pw38-xv9x-h8ch - CC BY 4.0

OSV / GitHub Advisory Database - PYSEC-2026-1565 - PyPA Advisory Database; CC BY 4.0

Vulnerability history: llama-index

All published vulnerabilities of this package (Python (PyPI)), newest first. Shows whether this is a one-off slip or a package that needs watching.

All (22) in the archive
Total entries
22
First → latest
08.2023 01.2026
Without a fix
2
Exploited in attacks
0

Severity breakdown

  • Critical 5
  • High 11
  • Medium 6

Disclosures per year

Severity:
  1. LlamaIndex (run-llama/llama_index) versions up to and including 0.12.2 contain an uncontrolled resource consumption vulnerability in the VannaPack VannaQueryEngine implementation. The custom_query() l
    High · 7.5 CVE-2024-58339 < 0.12.3 0.12.3
  2. LlamaIndex (run-llama/llama_index) versions up to and including 0.11.6 contain an unsafe deserialization vulnerability in BGEM3Index.load_from_disk() in llama_index/indices/managed/bge_m3/base.py. The
    High · 7.8 CVE-2024-14021 < 0.11.7 0.11.7
  3. llama-index has Insecure Temporary File
    High · 7.1 CVE-2025-7707 < 0.13.0 0.13.0
  4. LlamaIndex vulnerable to Path Traversal attack through its encode_image function
    High · 7.5 CVE-2025-6209 >= 0.12.27, < 0.12.41 0.12.41
  5. LlamaIndex has an XML Entity Expansion vulnerability in its sitemap parser
    High · 7.5 CVE-2025-3225 >= 0.12.21, < 0.12.29 0.12.29
  6. LlamaIndex is vulnerable to Path Traversal attack through its ObsidianReader class
    High · 7.5 CVE-2025-3046 >= 0.12.23, < 0.12.28 0.12.28
  7. llama_index vulnerable to SQL Injection
    Critical · 9.8 CVE-2025-1793 < 0.12.28 0.12.28
  8. LlamaIndex DuckDBVectorStore vulnerable to SQL injection through ref_doc_id
    Critical · 9.8 CVE-2025-1750 >= 0.12.19, < 0.12.21 0.12.21
  9. LLama-Index CLI OS command injection vulnerability
    High · 7.8 CVE-2025-1753 <= 0.12.20 no fix
  10. LlamaIndex Vulnerable to Denial of Service (DoS)
    High · 7.5 CVE-2025-1752 >= 0.12.15, < 0.12.21 0.12.21
  11. LlamaIndex vulnerable to Creation of Temporary File in Directory with Insecure Permissions
    High · 7.1 CVE-2024-12911 < 0.12.3 0.12.3
  12. LlamaIndex Improper Handling of Exceptional Conditions vulnerability
    High · 7.5 CVE-2024-12704 <= 0.12.5 no fix
  13. LlamaIndex includes an exec call for `import {cls_name}`
    Critical · 9.8 CVE-2024-45201 < 0.10.38 0.10.38
  14. RunGptLLM class in LlamaIndex has a command injection this entry
    High · 8.8 CVE-2024-4181 >= 0.9.47, < 0.10.13 0.10.13
  15. LlamaIndex (aka llama_index) through 0.9.34 allows SQL injection via the Text-to-SQL feature in NLSQLTableQueryEngine, SQLTableRetrieverQueryEngine, NLSQLRetriever, RetrieverQueryEngine, and PGVectorS
    Critical · 9.8 CVE-2024-23751 < 0.9.35 0.9.35
  16. An issue in llama_index v.0.7.13 and before allows a remote attacker to execute arbitrary code via the `exec` parameter in PandasQueryEngine function.
    Critical · 9.8 CVE-2023-39662 < 0.7.14 0.7.14

Know which vulnerabilities affect your application

With the DockRay integration and software inventory enabled, DockRay checks reported component versions against known vulnerabilities and notifies you about confirmed matches. Coverage depends on the supported technology and available advisory data.

Connect your application
DockRay newsletter

Application security in your inbox

Important vulnerabilities, actionable guidance and practical advice for your applications. Choose your technologies and follow what matters to your project.

  • A daily digest of new vulnerabilities, critical ones included.
  • Notifications about important updates.
  • Practical advice on application security and monitoring.

Leave empty to receive updates for every technology.

Privacy policy

Free to subscribe. Unsubscribe with one click in any email.