vLLM: Incomplete CVE-2025-62164 remediation can be bypassed by concurrent prompt parts

Executive SummaryThe follow-up protection for CVE-2025-62164 is incomplete at vLLM revision 26587f9519e22a5c4549ead7595ad9ca3229c4fd. It wraps serialized prompt-embedding reconstruction and dense conversion in torch.sparse.checksparsetensorinvariants, but PyTorch 2.11.0 implements that context wi...

PYSEC-2026-3938 vLLM: Incomplete CVE-2025-62164 remediation can be bypassed by concurrent prompt parts

Executive Summary The follow-up protection for CVE-2025-62164 is incomplete at vLLM revision 26587f9519e22a5c4549ead7595ad9ca3229c4fd. It wraps serialized prompt-embedding reconstruction and dense conversion in torch.sparse.checksparsetensorinvariants, but PyTorch 2.11.0 implements that context...

vLLM: Derender endpoints decode caller-supplied GenerateResponse token IDs without output bounds

SummaryThe /v1/completions/derender and /v1/chat/completions/derender endpoints accept caller-supplied GenerateResponse objects and postprocess every nested choices.tokenids list directly. Unlike the normal render/generate path, derender does not enforce model context length, resolved maxtokens,...

PYSEC-2026-3936 vLLM: Derender endpoints decode caller-supplied GenerateResponse token IDs without output bounds

Summary The /v1/completions/derender and /v1/chat/completions/derender endpoints accept caller-supplied GenerateResponse objects and postprocess every nested choices.tokenids list directly. Unlike the normal render/generate path, derender does not enforce model context length, resolved maxtokens,...

Omnigent: Shared Agent Bundle Overwrite Leads to Authenticated Runner RCE

SummaryAn authenticated user with edit access to their own session can overwrite a shared/template agent by uploading a full agent bundle through PUT /sessions/sessionid/agent.Shared/template agents are shown as not MCP-editable, but this upload path still accepts a replacement bundle. By adding ...

unstructured: Server-Side Request Forgery in the URL-based partitioning

SummaryServer-Side Request Forgery in unstructured. The url= argument of partition, partitionhtml, and partitionmd is fetched via requests.get with no host validation. The response body is returned as Element text, so this is a full-read SSRF — attackers reach loopback admin APIs, internal HTTP...

Omnigent: Unvalidated os_env.cwd in agent bundle yields arbitrary host filesystem access on runners without OMNIGENT_RUNNER_WORKSPACE

SummaryAn authenticated, non-admin user can obtain arbitrary host-filesystem read/write and host environment-secret disclosure on an Omnigent runner by uploading an agent bundle whose osenv.cwd points outside any intended workspace e.g. / or /home/. The cwd field is taken verbatim from the bundle...

PYSEC-2026-3930 unstructured: Server-Side Request Forgery in the URL-based partitioning

Summary Server-Side Request Forgery in unstructured. The url= argument of partition, partitionhtml, and partitionmd is fetched via requests.get with no host validation. The response body is returned as Element text, so this is a full-read SSRF — attackers reach loopback admin APIs, internal HTTP...

PYSEC-2026-3873 Omnigent Guardrail policy bypass: shell-command parser fails open in policies/builtins/_shell.py

Reporter: Aaron / Aeon — autonomous security agent https://github.com/aaronjmars/aeon Project: omnigent-ai/omnigent v0.1.0 Databricks — meta-harness running Claude Code / Codex / Pi "in check with policies and sandboxing" Component: omnigent/policies/builtins/shell.py shared parser → consumed by...

PYSEC-2026-3872 Omnigent: Uploaded Agent Bundle Allows Authenticated Runner RCE via Python Callable Tools

Summary An authenticated user can upload a crafted agent bundle that defines a server-side Python callable tool. The server validates the uploaded bundle, but it does not block dangerous callable: paths in untrusted user-provided agent configs. When the tool is invoked, the runner imports and...

PYSEC-2026-3874 Omnigent: Shared Agent Bundle Overwrite Leads to Authenticated Runner RCE

Summary An authenticated user with edit access to their own session can overwrite a shared/template agent by uploading a full agent bundle through PUT /sessions/sessionid/agent. Shared/template agents are shown as not MCP-editable, but this upload path still accepts a replacement bundle. By addin...

Omnigent Guardrail policy bypass: shell-command parser fails open in policies/builtins/_shell.py

Reporter: Aaron / Aeon — autonomous security agent https://github.com/aaronjmars/aeonProject: omnigent-ai/omnigent v0.1.0 Databricks — meta-harness running Claude Code / Codex / Pi "in check with policies and sandboxing"Component: omnigent/policies/builtins/shell.py shared parser → consumed by...

PYSEC-2026-3875 Omnigent: Unvalidated os_env.cwd in agent bundle yields arbitrary host filesystem access on runners without OMNIGENT_RUNNER_WORKSPACE

Summary An authenticated, non-admin user can obtain arbitrary host-filesystem read/write and host environment-secret disclosure on an Omnigent runner by uploading an agent bundle whose osenv.cwd points outside any intended workspace e.g. / or /home/. The cwd field is taken verbatim from the bundl...

Omnigent: Uploaded Agent Bundle Allows Authenticated Runner RCE via Python Callable Tools

SummaryAn authenticated user can upload a crafted agent bundle that defines a server-side Python callable tool. The server validates the uploaded bundle, but it does not block dangerous callable: paths in untrusted user-provided agent configs.When the tool is invoked, the runner imports and...

RestrictedPython guard hooks can be shadowed via positional-only arguments

ImpactRestrictedPython rewrites sensitive operations to go through guard hooks. Attribute access becomes getattrobj, name, item access becomes getitemobj, key, writes go through write, and print goes through print. The embedding application supplies these hooks to enforce its policy.Argument-name...

PYSEC-2026-3917 RestrictedPython guard hooks can be shadowed via positional-only arguments

Impact RestrictedPython rewrites sensitive operations to go through guard hooks. Attribute access becomes getattrobj, name, item access becomes getitemobj, key, writes go through write, and print goes through print. The embedding application supplies these hooks to enforce its policy. Argument-na...

asyncssh has SCP Path Traversal to Arbitrary File Write

| | ||---|---|| Product | asyncssh all versions through 2.23.0 || Related | CVE-2019-6111 same class in OpenSSH || Fix | AsyncSSH 2.23.1 |A malicious SSH server can write arbitrary files on the asyncssh SCP client's filesystem by sending filenames containing ../ traversal sequences. The SCP recei...

Trestle has Server-Side Template Injection (SSTI) via Recursive Template Re-evaluation of Untrusted Data

ImpactA Server-Side Template Injection SSTI vulnerability exists in multiple locations of trestle's Jinja2 rendering pipeline due to a systemic pattern: untrusted data is re-parsed as Jinja2 template source code without sandboxing. This advisory tracks the root cause across all affected code...

WebOb: Open redirect in Location header normalization via leading C0 control / space characters

SummaryThis is a third follow-up to CVE-2024-42353 / GHSA-mg3v-6m49-jhp3and CVE-2026-44889 / GHSA-fh3h-vg37-cc95.WebOb makes the Location header absolute when it serves a redirect. To stop arelative or protocol-relative target from redirecting users off-host, it checksthe value for a URI scheme a...

PYSEC-2026-3805 aiosmtplib: STARTTLS response injection

Impact When a connection is upgraded with STARTTLS, aiosmtplib reads the server's 220 go-ahead reply and immediately performs the TLS handshake without discarding any data still sitting in the receive buffer. Bytes the protocol read off the plaintext socket before the handshake survive across the...