UTC

litellm

45 published advisories; version 1.104.0rc2 or later fixes all but 13 of them.

Compromised releases

1.107.0, 1.82.7, 1.82.8 were published with malicious code. If any of them was installed, remove it, rotate every credential the environment could read, and pin dependencies by hash.

Hit by supply-chain campaigns: LiteLLM release pipeline compromise (1.107.0), LiteLLM PyPI compromise (1.82.7 and 1.82.8).

Registry
PyPI
Advisories
45
Latest fix
1.104.0rc2
Exploited in the wild
4
Advisories with no fix yet

13 advisories have no fixed release in any version yet. What can be done meanwhile:

General guidance from ShieldPi, not a vendor fix

  • Compromised release. Move to a release the advisory names as clean, and rotate every credential, token and key the environment that ran the affected release could read.GHSA-5mg7-485q-xm76
  • Command or code injection. Remove or disable the affected server or tool, or run it in a sandbox with no secrets in its environment.GHSA-53gh-p8jc-7rg8 · GHSA-7ggm-4rjg-594w
  • Server-side request forgery. Restrict its outbound network access to the hosts it needs, and keep internal addresses and cloud metadata endpoints out of its reach.GHSA-c693-x898-5g4h
  • Missing or bypassable authentication. Do not expose it on a network others can reach, and put authentication you control in front of it, such as an authenticating proxy or a VPN. Binding it to localhost is not enough on its own: a web page open in your browser can still reach it, so require a token as well.GHSA-j37q-q7p9-vpwm

Check your machine

Finds litellm in your agent and MCP configs, installed skills, global npm packages, the npx and uvx caches, uv and pipx tools, Claude Code and Cursor. It runs offline and sends nothing anywhere.

pip install "shieldpi>=0.10.0"
shieldpi agent-config . --installed
Downstream in the Official MCP Registry

3 registered MCP servers install an affected version today, by the dependency versions a fresh install resolves.

  • io.github.airas-org/airaslitellm@1.84.0
  • io.github.Brightdotdev/darwin-raglitellm@1.83.14
  • io.github.hzy9981/aviation-accident-analysislitellm@1.100.1
See the trace →
Provenance

Source: github.com/berriai/litellm, named in the package metadata (not verified)

3 of the last 10 releases of litellm were published with build provenance: a signed record of the repository and commit each was built from, which the registry checks before it accepts the release.

  • 1.110.0
  • 1.108.11
  • 1.108.10
  • 1.108.9
  • 1.106.10
  • 1.108.8
  • 1.108.7
  • 1.109.1
  • 1.108.6
  • 1.106.9

Green: published with provenance. Grey: without.

OpenSSF Scorecard has not scored this repository.

Provenance from deps.dev (CC BY 4.0); Scorecard from OpenSSF (CDLA-Permissive-2.0). As of 2026-11-18.

Advisories
AdvisorySeverityAffectedFixed inPublished
LiteLLM: Privilege Escalation to Proxy Admin via Cross-Domain Reuse of the Salt KeyCVE-2026-114612High>= 1.101.0, < 1.101.31.101.32026-10-27
LiteLLM: Privilege Escalation to Proxy Admin via Cross-Domain Reuse of the Salt KeyCVE-2026-114612High>= 1.102.0, < 1.102.21.102.22026-10-27
LiteLLM: Privilege Escalation to Proxy Admin via Cross-Domain Reuse of the Salt KeyCVE-2026-114612High>= 1.103.0, < 1.103.11.103.12026-10-27
LiteLLM: Privilege Escalation to Proxy Admin via Cross-Domain Reuse of the Salt KeyCVE-2026-114612High>= 1.104.0rc1, < 1.104.0rc21.104.0rc22026-10-27
LiteLLM: Privilege Escalation to Proxy Admin via Cross-Domain Reuse of the Salt KeyCVE-2026-114612High>= 1.91.0, < 1.100.41.100.42026-10-27
LiteLLM: internal_user_viewer arbitrary local file read via `vertex_ai_credentials`GHSA-g5ff-637f-6q2mHigh< 1.95.01.95.02026-10-27
LiteLLM vulnerable to reflected XSS in /sso/debug/callbackGHSA-hhww-mrg2-969hMedium>= 1.65.5, < 1.85.01.85.02026-10-27
LiteLLM: Authenticated SSRF and provider-credential exfiltration via unvalidated request-body routing parametersCVE-2026-84377Medium< 1.88.61.88.62026-09-30
LiteLLM: Authenticated SSRF and provider-credential exfiltration via unvalidated request-body routing parametersCVE-2026-84377Medium>= 1.89.0, < 1.89.71.89.72026-09-30
LiteLLM: Authenticated SSRF and provider-credential exfiltration via unvalidated request-body routing parametersCVE-2026-84377Medium>= 1.90.0, < 1.90.71.90.72026-09-30
LiteLLM: Authenticated SSRF and provider-credential exfiltration via unvalidated request-body routing parametersCVE-2026-84377Medium>= 1.91.0, < 1.91.51.91.52026-09-30
LiteLLM: Authenticated SSRF and provider-credential exfiltration via unvalidated request-body routing parametersCVE-2026-84377Medium>= 1.92.0, < 1.92.21.92.22026-09-30
LiteLLM: Authenticated SSRF and provider-credential exfiltration via unvalidated request-body routing parametersCVE-2026-84377Medium>= 1.93.0, < 1.93.21.93.22026-09-30
LiteLLM: Authenticated SSRF and provider-credential exfiltration via unvalidated request-body routing parametersCVE-2026-84377Medium>= 1.94.0, < 1.94.31.94.32026-09-30
LiteLLM: Authenticated SSRF and provider-credential exfiltration via unvalidated request-body routing parametersCVE-2026-84377Medium>= 1.95.0, < 1.95.11.95.12026-09-30
LiteLLM: Authenticated SSRF and provider-credential exfiltration via unvalidated request-body routing parametersCVE-2026-84377Medium>= 1.96.0, < 1.96.21.96.22026-09-30
LiteLLM Proxy has server-side request forgery via the `user_config` request parameterCVE-2026-59823Medium<= 1.83.81.83.92026-09-17
LiteLLM vulnerable to server-side template injection in the /prompts/test endpointCVE-2026-37004Critical< 1.83.71.83.72026-08-27
LiteLLM: Local file read via request-supplied OIDC file referencesCVE-2026-59819Low< 1.83.101.83.102026-07-22
LiteLLM: Arbitrary file write via path traversal in Skills archive extractionCVE-2026-59820Medium< 1.83.71.83.72026-07-22
LiteLLM: Custom Code Guardrails production endpoints bypass code safety checksCVE-2026-59821Low< 1.82.01.82.02026-07-22
LiteLLM: MCP Authentication Bypass via OAuth2 Passthrough FallbackCVE-2026-59822Exploited in the wildHigh< 1.84.01.84.02026-07-22
Malicious code in litellm (PyPI)GHSA-92x9-889m-jgmwCritical1.107.0Any other release2026-07-21
Malicious code in litellm (PyPI)GHSA-92x9-889m-jgmwCritical1.82.7Any other release2026-07-21
Malicious code in litellm (PyPI)GHSA-92x9-889m-jgmwCritical1.82.8Any other release2026-07-21
LiteLLM: MCP Proxy Has Improper AuthenticationCVE-2026-12773Medium< 1.84.01.84.02026-06-21
LiteLLM: Admin Key Handler Has Improper AuthorizationCVE-2026-12770Low<= 1.63.1No fix yet2026-06-21
BerriAI litellm has SSRF via Unvalidated spec_path URL in MCP OpenAPI Spec LoaderCVE-2026-12798Low<= 1.82.2No fix yet2026-06-21
LiteLLM: SSO Debug Flow Has Improper AuthenticationCVE-2026-12795Medium<= 1.82.2No fix yet2026-06-21
BerriAI litellm: UI User Enumeration leads to System-Wide Information DisclosureCVE-2026-12799Low<= 1.82.2No fix yet2026-06-21
LiteLLM: PROXY_ADMIN database API Key Generator Has Insufficient Session ExpirationCVE-2026-12772Low<= 1.82.2No fix yet2026-06-21
BerriAI litellm has Security Feature Bypass in BannedKeywords and AzureContentSafety Guardrails via call_type Mismatch on Async EndpointsCVE-2026-12797Low<= 1.82.5No fix yet2026-06-21
LiteLLM: M2M JWT Handler Has Improper AuthorizationCVE-2026-12771Low<= 1.82.2No fix yet2026-06-21
BerriAI litellm: SSO Login Does Not Invalidate Previous UI Session TokensCVE-2026-12796Low<= 1.82.2No fix yet2026-06-21
LiteLLM: Authentication Bypass via Host Header InjectionCVE-2026-49468Critical< 1.84.01.84.02026-06-16
LiteLLM allows an authenticated internal_user to create API keys with access to routes that their role does not permitCVE-2026-47101High< 1.83.141.83.142026-05-21
LiteLLM allows a user to modify their own user_role via the /user/update endpointCVE-2026-47102High< 1.83.101.83.102026-05-21
LiteLLM has a sandbox escape in custom-code guardrailCVE-2026-40217Exploited in the wildHigh>= 1.81.8, < 1.83.101.83.102026-05-11
LiteLLM: Authenticated command execution via MCP stdio test endpointsCVE-2026-42271Exploited in the wildHigh>= 1.74.2, < 1.83.71.83.72026-04-25
LiteLLM has SQL Injection in Proxy API key verificationCVE-2026-42208Exploited in the wildCritical>= 1.81.16, < 1.83.71.83.72026-04-24
LiteLLM: Server-Side Template Injection in /prompts/test endpointCVE-2026-42203High>= 1.80.5, < 1.83.71.83.72026-04-24
LiteLLM: Password hash exposure and pass-the-hash authentication bypassGHSA-69x8-hrgq-fjj8High< 1.83.01.83.02026-04-08
LiteLLM: Privilege escalation via unrestricted proxy configuration endpointCVE-2026-35029High< 1.83.01.83.02026-04-03
LiteLLM: Authentication bypass via OIDC userinfo cache key collisionCVE-2026-35030Critical< 1.83.01.83.02026-04-03
Two LiteLLM versions published containing credential harvesting malwareGHSA-5mg7-485q-xm76Critical>= 1.82.7, <= 1.82.8No fix yet2026-03-25
LiteLLM Vulnerable to Remote Code Execution (RCE)CVE-2024-6825High>= 1.40.3.dev2, <= 1.40.12No fix yet2025-03-20
LiteLLM Has a Leakage of Langfuse API KeysCVE-2025-0330High<= 1.52.1No fix yet2025-03-20
LiteLLM Vulnerable to Denial of Service (DoS) via Crafted HTTP RequestCVE-2024-8984High< 1.56.21.56.22025-03-20
LiteLLM Has an Improper Authorization VulnerabilityCVE-2025-0628High< 1.61.151.61.152025-03-20
LiteLLM Reveals Portion of API Key via a Logging FileCVE-2024-9606High< 1.44.121.44.122025-03-20
LiteLLM Vulnerable to Denial of Service (DoS)CVE-2024-10188High< 1.53.1.dev11.53.1.dev12025-03-20
LiteLLM Server-Side Request Forgery (SSRF) vulnerabilityCVE-2024-6587High< 1.44.81.44.82024-09-13
litellm vulnerable to remote code execution based on using eval unsafelyCVE-2024-5751Critical< 1.40.161.40.162024-06-27
litellm vulnerable to improper access control in team managementCVE-2024-5710Medium< 1.40.151.40.152024-06-27
Arbitrary file deletion in litellmCVE-2024-4888High< 1.35.361.35.362024-06-06
SQL injection in litellmCVE-2024-4890Medium<= 1.27.14No fix yet2024-06-06
SQL injection in litellmCVE-2024-5225Medium< 1.40.01.40.02024-06-06
litellm passes untrusted data to `eval` function without sanitizationCVE-2024-4264High<= 1.28.11No fix yet2024-05-18
LiteLLM has Server-Side Template Injection vulnerability in /completions endpointCVE-2024-2952Critical< 1.34.421.34.422024-04-10

Sources: GitHub Advisory Database, reviewed and malware advisories (CC BY 4.0, https://github.com/advisories); advisories published by Cursor (https://github.com/cursor/cursor/security/advisories); advisories published by OpenClaw (https://github.com/openclaw/openclaw/security/advisories); advisories published by vLLM (https://github.com/vllm-project/vllm/security/advisories); advisories published by Open WebUI (https://github.com/open-webui/open-webui/security/advisories); advisories published by n8n (https://github.com/n8n-io/n8n/security/advisories); advisories published by Flowise (https://github.com/FlowiseAI/Flowise/security/advisories); advisories published by Langflow (https://github.com/langflow-ai/langflow/security/advisories); advisories published by the MCP Python SDK project (https://github.com/modelcontextprotocol/python-sdk/security/advisories); advisories published by MLflow (https://github.com/mlflow/mlflow/security/advisories); CISA Known Exploited Vulnerabilities catalog (CC0). Data as of 2026-11-22. ShieldMap describes how each problem works and how to fix it; it does not publish exploit code. Report a correction to [email protected].