GHSA-JP94-MMR3-2PVV
Embedding `extra_headers` and `extra_body` settings are recorded in traces
Texto original en inglés · Ver en español
CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:N/A:N
Fuente: GitHub
Ciclo de vida de la vulnerabilidad
| Fecha | Fase | Evento | Detalle |
|---|---|---|---|
| Divulgación | Publicada | GHSA-JP94-MMR3-2PVV divulgada · CVSS 6.5 (GitHub) | |
| Remediación | Parche 2.66.0 | pydantic-ai · 0 d desde divulgación | |
| Monitoreo | Detectada | Ingresó al índice | |
| Monitoreo | Verificada | Última verificación contra las fuentes |
Summary
With OpenTelemetry instrumentation enabled for embedding models, embedding requests record the `extra_headers` and `extra_body` embedding settings in their trace spans. Credentials passed through those settings can be read by anyone with access to the exported traces.
Details
Instrumented embedding models record the request's embedding settings on each embeddings span, including `extra_headers` and `extra_body` in full. Neither `include_content=False` nor `include_model_request_parameters=False` removes them. Chat model request spans and image generation spans do not record these settings.
You are affected if you instrument embedding models and pass authentication headers, API keys, or other secrets through `EmbeddingSettings.extra_headers` or `extra_body`, for example credentials for an LLM gateway or proxy. The assumed attacker is anyone who can read the application's exported traces but is not entitled to those credentials. Using a leaked credential against the upstream service is a follow-on attack.
Mitigation
Upgrade `pydantic-ai` and `pydantic-ai-slim` to a patched version on the maintained v2 or v1 line.
Until then, keep credentials out of embedding `extra_headers` and `extra_body` while instrumentation is enabled, for example by configuring them on the provider's client, or pass `instrument=False` to the `Embedder`. Restrict access to historical traces and rotate credentials those traces exposed.
Desglose de métricas
Rangos afectados
Estado del parche por paquete
| Paquete / producto | Estado | Corregida en | Desde divulgación | Última versión |
|---|---|---|---|---|
pypipydantic-ai |
Corregida | 2.66.0 | 0 d | 2.68.0 no afectada |
pypipydantic-ai-slim |
Corregida | 2.66.0 | 0 d | 2.71.0 no afectada |
2.66.0Arsenal público
Sin exploits, módulos ni PoC públicos conocidos a la fecha de verificación.
https://github.com/pydantic/pydantic-ai/security/advisories/GHSA-jp94-mmr3-2pvvLas referencias externas se muestran para copiar: cve.llama.ec no lo redirige a otros sitios.
