These pages are published after PatchSiren validates generated defensive summaries against stored public CVE and source evidence.
CVE-2026-78205 is a Server-Side Request Forgery (SSRF) vulnerability in BentoML, a framework for building and deploying machine learning models. The vulnerability arises from the framework's incomplete safeguard against outbound connections to private IP addresses, specifically allowing requests to the RFC 6598 shared address space (100.64.0.0/10). This oversight enables unauthenticated attackers to make [truncated]
CVE-2026-15035 is a low-severity vulnerability in Bentoml OpenLLM 0.6.30, affecting the async_run_command function in src/openllm/common.py, which is part of the Model Repository Directory Name Handler component. An attacker must have local access to exploit this vulnerability. The exploit has been made public, and the project has been informed but has not yet responded. The vulnerability allows for comma [truncated]
BentoML versions prior to 1.4.39 contain a template injection vulnerability in the container build pipeline. The Jinja2 template at `src/bentoml/_internal/container/frontend/dockerfile/templates/base_v2.j2` interpolates the `docker.base_image` configuration value without escaping, newline filtering, or validation. A maliciously crafted `bento.yaml` file with a multi-line `docker.base_image` value can inje [truncated]
A symlink-following vulnerability in BentoML's build packaging workflow allows attackers to exfiltrate local files from build hosts into Bento artifacts. When building untrusted repositories, attacker-controlled symlinks are dereferenced and their target file contents are copied into the generated artifact. This exposes sensitive files including cloud credentials, SSH keys, API tokens, and environment con [truncated]