GHSA-824W-X939-6CMC
Vulnerability from github – Published: 2026-07-15 21:59 – Updated: 2026-07-15 21:59Impact
The /internal/object_storage endpoint accepts a caller-supplied JSON storage_path parameter that dynamically overrides the TensorZero [object_storage] configuration.
By abusing the filesystem storage type, a caller can read arbitrary files from the gateway filesystem, including files that may contain sensitive credentials. Similarly, by abusing the s3_compatible storage type, the caller can coerce the gateway into making outbound object storage requests to attacker-chosen internal/cloud-metadata endpoints.
This vulnerability only applies when the gateway can be accessed by untrusted callers. If a developer's TensorZero deployment has authentication enabled, only authenticated callers can exploit this vulnerability. If a developer's deployment has authentication disabled, any caller can exploit this vulnerability.
Remediation
The vulnerability has been patched in version 2026.6.0. See PR #7527.
Workarounds
If developers are unable to upgrade a gateway that is exposed to untrusted callers, please block external access to the /internal/object_storage endpoint.
{
"affected": [
{
"package": {
"ecosystem": "PyPI",
"name": "tensorzero"
},
"ranges": [
{
"events": [
{
"introduced": "0"
},
{
"fixed": "2026.6.0"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [
"CVE-2026-54457"
],
"database_specific": {
"cwe_ids": [
"CWE-552",
"CWE-918"
],
"github_reviewed": true,
"github_reviewed_at": "2026-07-15T21:59:40Z",
"nvd_published_at": null,
"severity": "HIGH"
},
"details": "### Impact\n\nThe `/internal/object_storage` endpoint accepts a caller-supplied JSON `storage_path` parameter that dynamically overrides the TensorZero `[object_storage]` configuration.\n\nBy abusing the `filesystem` storage type, a caller can read arbitrary files from the gateway filesystem, including files that may contain sensitive credentials. Similarly, by abusing the `s3_compatible` storage type, the caller can coerce the gateway into making outbound object storage requests to attacker-chosen internal/cloud-metadata endpoints.\n\nThis vulnerability only applies when the gateway can be accessed by untrusted callers. If a developer\u0027s TensorZero deployment has authentication enabled, only authenticated callers can exploit this vulnerability. If a developer\u0027s deployment has authentication disabled, any caller can exploit this vulnerability.\n\n### Remediation\n\nThe vulnerability has been patched in version `2026.6.0`. See PR #7527.\n\n### Workarounds\n\nIf developers are unable to upgrade a gateway that is exposed to untrusted callers, please block external access to the `/internal/object_storage` endpoint.",
"id": "GHSA-824w-x939-6cmc",
"modified": "2026-07-15T21:59:40Z",
"published": "2026-07-15T21:59:40Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/tensorzero/tensorzero/security/advisories/GHSA-824w-x939-6cmc"
},
{
"type": "WEB",
"url": "https://github.com/tensorzero/tensorzero/commit/0abbc838bae3394fe7491dad7009670d4e3b6cf8"
},
{
"type": "PACKAGE",
"url": "https://github.com/tensorzero/tensorzero"
},
{
"type": "WEB",
"url": "https://github.com/tensorzero/tensorzero/releases/tag/2026.6.0"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:C/C:H/I:N/A:N",
"type": "CVSS_V3"
}
],
"summary": "TensorZero Gateway: Arbitrary file read and SSRF in internal object storage endpoint"
}
Sightings
| Author | Source | Type | Date | Other |
|---|
Nomenclature
- Seen: The vulnerability was mentioned, discussed, or observed by the user.
- Confirmed: The vulnerability has been validated from an analyst's perspective.
- Published Proof of Concept: A public proof of concept is available for this vulnerability.
- Exploited: The vulnerability was observed as exploited by the user who reported the sighting.
- Patched: The vulnerability was observed as successfully patched by the user who reported the sighting.
- Not exploited: The vulnerability was not observed as exploited by the user who reported the sighting.
- Not confirmed: The user expressed doubt about the validity of the vulnerability.
- Not patched: The vulnerability was not observed as successfully patched by the user who reported the sighting.