{"id":"CVE-2026-42203","aliases":["EUVD-2026-28502","GHSA-xqmj-j6mv-4862","PYSEC-2026-2602"],"status":"published","title":"LiteLLM: Server-Side Template Injection in /prompts/test endpoint","description":"LiteLLM is a proxy server (AI Gateway) to call LLM APIs in OpenAI (or native) format. From version 1.80.5 to before version 1.83.7, the POST /prompts/test endpoint accepted user-supplied prompt templates and rendered them without sandboxing. A crafted template could run arbitrary code inside the LiteLLM Proxy process. The endpoint only checks that the caller presents a valid proxy API key, so any authenticated user could reach it. Depending on how the proxy is deployed, this could expose secrets in the process environment (such as provider API keys or database credentials) and allow commands to be run on the host. This issue has been patched in version 1.83.7.","severity":{"version":"4.0","score":8.6,"vector":"CVSS:4.0/AV:N/AC:L/AT:N/PR:L/UI:N/VC:H/VI:H/VA:N/SC:N/SI:N/SA:N","level":"high","assessor":"cna","from":"cve"},"epss":{"day":"2026-11-11","score":0.00373,"percentile":0.3171,"model":"v2026.06.15"},"facts":{"in_kev":false,"known_exploited":false,"public_exploit":true,"fix_available":true},"cwes":["CWE-1336","CWE-94"],"published":"2026-04-24T16:02:42Z","modified":"2026-10-28T02:30:29Z","exploits":[{"source":"pocgithub","id":"Astianjy/CVE-2026-42203","title":"Astianjy/CVE-2026-42203","date":"2026-05-15","url":"https://github.com/Astianjy/CVE-2026-42203"}],"fixes":[{"ecosystem":"PyPI","package":"litellm","fixed_in":["1.83.7"],"sources":["ghsa","osv"]}],"affected":[{"source":"ghsa","ecosystem":"PyPI","package":"litellm","ranges":[{"type":"ECOSYSTEM","events":[{"introduced":"1.80.5"},{"fixed":"1.83.7"}]}]},{"source":"osv","ecosystem":"PyPI","package":"litellm","ranges":[{"type":"ECOSYSTEM","events":[{"introduced":"1.80.5"},{"fixed":"1.83.7"}]}],"versions":["1.80.10","1.80.11","1.80.12","1.80.13","1.80.15","1.80.16","1.80.17","1.80.5","1.80.6","1.80.7","1.80.8","1.80.9","1.81.0","1.81.1","1.81.10","1.81.11","1.81.12","1.81.13","1.81.14","1.81.15","1.81.16","1.81.3","1.81.4","1.81.5","1.81.6","1.81.7","1.81.8","1.81.9","1.81.9.dev1","1.82.0","1.82.1","1.82.2","1.82.3","1.82.4","1.82.5","1.82.6","1.83.0","1.83.1","1.83.2","1.83.3","1.83.4","1.83.5","1.83.6"]},{"source":"osv","ecosystem":"PyPI","package":"litellm","ranges":[{"type":"ECOSYSTEM","events":[{"introduced":"1.80.5"},{"fixed":"1.83.7"}]}],"versions":["1.80.10","1.80.11","1.80.12","1.80.13","1.80.15","1.80.16","1.80.17","1.80.5","1.80.6","1.80.7","1.80.8","1.80.9","1.81.0","1.81.1","1.81.10","1.81.11","1.81.12","1.81.13","1.81.14","1.81.15","1.81.16","1.81.3","1.81.4","1.81.5","1.81.6","1.81.7","1.81.8","1.81.9","1.81.9.dev1","1.82.0","1.82.1","1.82.2","1.82.3","1.82.4","1.82.5","1.82.6","1.83.0","1.83.1","1.83.2","1.83.3","1.83.4","1.83.5","1.83.6"]}],"products":[{"vendor":"berriai","product":"litellm"},{"vendor":"litellm","product":"litellm"},{"vendor":"red hat","product":"exploit intelligence"},{"vendor":"red hat","product":"red hat ansible automation platform 2"},{"vendor":"red hat","product":"red hat openshift ai (rhoai)"}],"references":[{"url":"https://github.com/BerriAI/litellm/security/advisories/GHSA-xqmj-j6mv-4862","tags":["x_refsource_CONFIRM","Mitigation","Patch","Vendor Advisory","advisory","web"]},{"url":"https://github.com/BerriAI/litellm/releases/tag/v1.83.7-stable","tags":["x_refsource_MISC","Product","Release Notes","web"]},{"url":"https://access.redhat.com/security/cve/CVE-2026-42203","tags":["vdb-entry","x_refsource_REDHAT","advisory"]},{"url":"https://bugzilla.redhat.com/show_bug.cgi?id=2467917","tags":["issue-tracking","x_refsource_REDHAT","report"]},{"url":"https://security.access.redhat.com/data/csaf/v2/vex/2026/cve-2026-42203.json","tags":["x_sadp-csaf-vex","web"]},{"url":"https://nvd.nist.gov/vuln/detail/CVE-2026-42203","tags":["advisory"]},{"url":"https://github.com/advisories/GHSA-xqmj-j6mv-4862","tags":["advisory"]},{"url":"https://github.com/CVEProject/cvelistV5/tree/main/cves/2026/42xxx/CVE-2026-42203.json","tags":["advisory"]},{"url":"https://github.com/BerriAI/litellm","tags":["package"]},{"url":"https://pypi.org/project/litellm","tags":["package"]}],"sources":{"cve":[{"id":"CVE-2026-42203","title":"LiteLLM: Server-Side Template Injection in /prompts/test endpoint","description":"LiteLLM is a proxy server (AI Gateway) to call LLM APIs in OpenAI (or native) format. From version 1.80.5 to before version 1.83.7, the POST /prompts/test endpoint accepted user-supplied prompt templates and rendered them without sandboxing. A crafted template could run arbitrary code inside the LiteLLM Proxy process. The endpoint only checks that the caller presents a valid proxy API key, so any authenticated user could reach it. Depending on how the proxy is deployed, this could expose secrets in the process environment (such as provider API keys or database credentials) and allow commands to be run on the host. This issue has been patched in version 1.83.7.","status":"published","published":"2026-05-08T03:36:58.648Z","modified":"2026-07-15T00:58:06.313Z","cvss":[{"version":"4.0","score":8.6,"vector":"CVSS:4.0/AV:N/AC:L/AT:N/PR:L/UI:N/VC:H/VI:H/VA:N/SC:N/SI:N/SA:N","level":"high","assessor":"cna"},{"version":"3.1","score":8.8,"vector":"CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H","level":"high","assessor":"adp"}],"cwes":["CWE-1336","CWE-94"],"updated_at":"2026-11-02T13:40:35.004553Z"}],"euvd":[{"id":"EUVD-2026-28502","description":"LiteLLM is a proxy server (AI Gateway) to call LLM APIs in OpenAI (or native) format. From version 1.80.5 to before version 1.83.7, the POST /prompts/test endpoint accepted user-supplied prompt templates and rendered them without sandboxing. A crafted template could run arbitrary code inside the LiteLLM Proxy process. The endpoint only checks that the caller presents a valid proxy API key, so any authenticated user could reach it. Depending on how the proxy is deployed, this could expose secrets in the process environment (such as provider API keys or database credentials) and allow commands to be run on the host. This issue has been patched in version 1.83.7.","status":"published","published":"2026-05-08T03:36:58Z","modified":"2026-07-15T00:58:06Z","cvss":[{"version":"4.0","score":8.6,"vector":"CVSS:4.0/AV:N/AC:L/AT:N/PR:L/UI:N/VC:H/VI:H/VA:N/SC:N/SI:N/SA:N","level":"high"}],"updated_at":"2026-11-02T18:19:56.918546Z"}],"ghsa":[{"id":"GHSA-xqmj-j6mv-4862","title":"LiteLLM: Server-Side Template Injection in /prompts/test endpoint","description":"### Impact\nThe `POST /prompts/test` endpoint accepted user-supplied prompt templates and rendered them without sandboxing. A crafted template could run arbitrary code inside the LiteLLM Proxy process.\n\nThe endpoint only checks that the caller presents a valid proxy API key, so any authenticated user could reach it. Depending on how the proxy is deployed, this could expose secrets in the process environment (such as provider API keys or database credentials) and allow commands to be run on the host.\n\nProxy deployments running an affected version are in scope.\n\n### Patches\nThe issue is fixed in **`1.83.7-stable`**. The fix switches the prompt template renderer to a sandboxed environment that blocks the attributes this attack relies on.\n\nLiteLLM recommends upgrading to `1.83.7-stable` or later.\n\n### Workarounds\nIf upgrading is not immediately possible:\n\n1. Block `POST /prompts/test` at your reverse proxy or API gateway.\n2. Review and rotate API keys that should not have access to prompt management routes.","status":"published","published":"2026-04-24T16:02:42Z","modified":"2026-05-12T13:27:02Z","cvss":[{"version":"4.0","score":8.6,"vector":"CVSS:4.0/AV:N/AC:L/AT:N/PR:L/UI:N/VC:H/VI:H/VA:N/SC:N/SI:N/SA:N","level":"high"}],"cwes":["CWE-1336"],"updated_at":"2026-11-02T14:40:05.816188Z"}],"nvd":[{"id":"CVE-2026-42203","description":"LiteLLM is a proxy server (AI Gateway) to call LLM APIs in OpenAI (or native) format. From version 1.80.5 to before version 1.83.7, the POST /prompts/test endpoint accepted user-supplied prompt templates and rendered them without sandboxing. A crafted template could run arbitrary code inside the LiteLLM Proxy process. The endpoint only checks that the caller presents a valid proxy API key, so any authenticated user could reach it. Depending on how the proxy is deployed, this could expose secrets in the process environment (such as provider API keys or database credentials) and allow commands to be run on the host. This issue has been patched in version 1.83.7.","status":"published","published":"2026-05-08T04:16:19.45Z","modified":"2026-10-26T20:10:00.16Z","cvss":[{"version":"4.0","score":8.6,"vector":"CVSS:4.0/AV:N/AC:L/AT:N/PR:L/UI:N/VC:H/VI:H/VA:N/SC:N/SI:N/SA:N/E:X/CR:X/IR:X/AR:X/MAV:X/MAC:X/MAT:X/MPR:X/MUI:X/MVC:X/MVI:X/MVA:X/MSC:X/MSI:X/MSA:X/S:X/AU:X/R:X/V:X/RE:X/U:X","level":"high","assessor":"secondary"},{"version":"3.1","score":8.8,"vector":"CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H","level":"high","assessor":"primary"},{"version":"3.1","score":8.8,"vector":"CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H","level":"high","assessor":"secondary"}],"cwes":["CWE-1336","CWE-94"],"updated_at":"2026-11-02T14:12:17.842772Z"}],"osv":[{"id":"CVE-2026-42203","title":"LiteLLM: Server-Side Template Injection in /prompts/test endpoint","description":"LiteLLM is a proxy server (AI Gateway) to call LLM APIs in OpenAI (or native) format. From version 1.80.5 to before version 1.83.7, the POST /prompts/test endpoint accepted user-supplied prompt templates and rendered them without sandboxing. A crafted template could run arbitrary code inside the LiteLLM Proxy process. The endpoint only checks that the caller presents a valid proxy API key, so any authenticated user could reach it. Depending on how the proxy is deployed, this could expose secrets in the process environment (such as provider API keys or database credentials) and allow commands to be run on the host. This issue has been patched in version 1.83.7.","status":"published","published":"2026-05-08T03:36:58.648Z","modified":"2026-10-28T02:30:29.932349628Z","cvss":[{"version":"4.0","score":8.6,"vector":"CVSS:4.0/AV:N/AC:L/AT:N/PR:L/UI:N/VC:H/VI:H/VA:N/SC:N/SI:N/SA:N","level":"high"}],"cwes":["CWE-1336"],"updated_at":"2026-11-02T15:45:04.117819Z"},{"id":"GHSA-xqmj-j6mv-4862","title":"LiteLLM: Server-Side Template Injection in /prompts/test endpoint","description":"### Impact\nThe `POST /prompts/test` endpoint accepted user-supplied prompt templates and rendered them without sandboxing. A crafted template could run arbitrary code inside the LiteLLM Proxy process.\n\nThe endpoint only checks that the caller presents a valid proxy API key, so any authenticated user could reach it. Depending on how the proxy is deployed, this could expose secrets in the process environment (such as provider API keys or database credentials) and allow commands to be run on the host.\n\nProxy deployments running an affected version are in scope.\n\n### Patches\nThe issue is fixed in **`1.83.7-stable`**. The fix switches the prompt template renderer to a sandboxed environment that blocks the attributes this attack relies on.\n\nLiteLLM recommends upgrading to `1.83.7-stable` or later.\n\n### Workarounds\nIf upgrading is not immediately possible:\n\n1. Block `POST /prompts/test` at your reverse proxy or API gateway.\n2. Review and rotate API keys that should not have access to prompt management routes.","status":"published","published":"2026-04-24T16:02:42Z","modified":"2026-09-10T03:50:45.007056853Z","cvss":[{"version":"4.0","score":8.6,"vector":"CVSS:4.0/AV:N/AC:L/AT:N/PR:L/UI:N/VC:H/VI:H/VA:N/SC:N/SI:N/SA:N","level":"high"}],"cwes":["CWE-1336"],"updated_at":"2026-11-02T15:47:00.198893Z"},{"id":"PYSEC-2026-2602","title":"LiteLLM: Server-Side Template Injection in /prompts/test endpoint","description":"### Impact\nThe `POST /prompts/test` endpoint accepted user-supplied prompt templates and rendered them without sandboxing. A crafted template could run arbitrary code inside the LiteLLM Proxy process.\n\nThe endpoint only checks that the caller presents a valid proxy API key, so any authenticated user could reach it. Depending on how the proxy is deployed, this could expose secrets in the process environment (such as provider API keys or database credentials) and allow commands to be run on the host.\n\nProxy deployments running an affected version are in scope.\n\n### Patches\nThe issue is fixed in **`1.83.7-stable`**. The fix switches the prompt template renderer to a sandboxed environment that blocks the attributes this attack relies on.\n\nLiteLLM recommends upgrading to `1.83.7-stable` or later.\n\n### Workarounds\nIf upgrading is not immediately possible:\n\n1. Block `POST /prompts/test` at your reverse proxy or API gateway.\n2. Review and rotate API keys that should not have access to prompt management routes.","status":"published","published":"2026-07-13T15:02:53.26715Z","modified":"2026-07-13T16:32:04.546343789Z","cvss":[{"version":"4.0","score":8.6,"vector":"CVSS:4.0/AV:N/AC:L/AT:N/PR:L/UI:N/VC:H/VI:H/VA:N/SC:N/SI:N/SA:N","level":"high"}],"updated_at":"2026-11-02T15:52:55.961245Z"}]}}