CVE-2026-53998: Phoenix: Format String Injection in applyChatTemplate
Exploitation: Public exploit · CVSS 10.0 Critical · Unauthenticated · Published 2026-11-16
Description
Phoenix is an AI observability and evaluation platform. From 13.0.0 until 15.7.0, the applyChatTemplate GraphQL query passed attacker-controlled F-string templates to FStringTemplateFormatter._format, while DotDict.__getattr__ did not prevent traversal through Python private attributes. A crafted field could traverse object internals such as __class__, __init__, and __globals__ to read process environment variables and internal module state. The endpoint is unauthenticated on default installations, and any account can reach it when authentication is enabled. Exposure of PHOENIX_SECRET can allow administrative session-token forgery and decryption of credentials protected by keys derived from that secret. This issue is fixed in version 15.7.0.
Enclave's analysis
Arize Phoenix 13.0.0 to before 15.7.0 renders attacker-supplied F-string templates in the applyChatTemplate GraphQL query. Dunder traversal leaks the server's environment variables, including PHOENIX_SECRET. One HTTP request does it, with no login on default installs and any account when auth is on. A public PoC exists, but no exploitation has been seen. Upgrade to 15.7.0.
Fix
Upgrade Arize Phoenix (arize-phoenix) to 15.7.0. No fixed version is listed for phoenix; follow the vendor advisory.
Affected versions
| Product | Vulnerable | Fixed in |
|---|---|---|
| phoenix | >= 13.0.0, < 15.7.0 | Not listed |
| Arize Phoenix (arize-phoenix) | >= 13.0.0, < 15.7.0 | 15.7.0 |
Affected packages
- pypi:arize-phoenix