AEGIS: From Clues to Verdicts -- Graph-Guided Deep Vulnerability Reasoning Via Dialectics and Meta-Auditing

Large Language Models LLMs are increasingly adopted for vulnerability detection, yet their reasoning remains fundamentally unsound. We identify a root cause shared by both major mitigation paradigms agent-based debate and retrieval augmentation: reasoning in an ungrounded deliberative space that...

When Scanners Lie: Evaluator Instability in LLM Red-Teaming

Automated LLM vulnerability scanners are increasingly used to assess security risks by measuring different attack type success rates ASR. Yet the validity of these measurements hinges on an often-overlooked component: the evaluator who determines whether an attack has succeeded. In this study, we...

GHSA-MHG6-2Q2V-9H2C sigstore-ruby verifier returns success for DSSE bundles with mismatched in-toto subject digest

Summary Sigstore::Verifierverify does not propagate the VerificationFailure returned by verifyintoto when the artifact digest does not match the digest in the in-toto attestation subject. As a result, verification of DSSE bundles containing in-toto statements returns VerificationSuccess regardles...

sigstore-ruby verifier returns success for DSSE bundles with mismatched in-toto subject digest

Summary Sigstore::Verifierverify does not propagate the VerificationFailure returned by verifyintoto when the artifact digest does not match the digest in the in-toto attestation subject. As a result, verification of DSSE bundles containing in-toto statements returns VerificationSuccess regardles...

sigstore-ruby verifier returns success for DSSE bundles with mismatched in-toto subject digest

Summary Sigstore::Verifierverify does not propagate the VerificationFailure returned by verifyintoto when the artifact digest does not match the digest in the in-toto attestation subject. As a result, verification of DSSE bundles containing in-toto statements returns VerificationSuccess regardles...

CVE-2026-31830

Summary: CVE-2026-31830 affects the sigstore-ruby project before version 0.2.3. The bug is in Sigstore::Verifier#verify, which fails to propagate the VerificationFailure returned by verify_in_toto when the artifact digest does not match the digest in the in-toto attestation subject. As a result, ...

CVE-2026-31830 sigstore-ruby verifier returns success for DSSE bundles with mismatched in-toto subject digest

sigstore-ruby is a pure Ruby implementation of the sigstore verify command from the sigstore/cosign project. Prior to 0.2.3, Sigstore::Verifierverify does not propagate the VerificationFailure returned by verifyintoto when the artifact digest does not match the digest in the in-toto attestation...

CVE-2026-31830 sigstore-ruby verifier returns success for DSSE bundles with mismatched in-toto subject digest

sigstore-ruby is a pure Ruby implementation of the sigstore verify command from the sigstore/cosign project. Prior to 0.2.3, Sigstore::Verifierverify does not propagate the VerificationFailure returned by verifyintoto when the artifact digest does not match the digest in the in-toto attestation...

Unicorn 安全漏洞

Unicorn is an open-source verifier developed by the World Wide Web Consortium. It helps people improve the quality of web pages by performing various checks. Versions of Unicorn prior to 0.67.0 contained security vulnerabilities; these vulnerabilities stemmed from lack of access control checks...

OpenClaw: macOS beta onboarding exposed PKCE verifier via OAuth state

Summary The affected surface is the OpenClaw macOS app onboarding flow, and the macOS app is currently in beta. In that beta onboarding flow, Anthropic OAuth used the PKCE codeverifier value as OAuth state, exposing that secret in front-channel URL state. Affected Packages / Versions - Package:...

GHSA-6G25-PC82-VFWP OpenClaw: macOS beta onboarding exposed PKCE verifier via OAuth state

Summary The affected surface is the OpenClaw macOS app onboarding flow, and the macOS app is currently in beta. In that beta onboarding flow, Anthropic OAuth used the PKCE codeverifier value as OAuth state, exposing that secret in front-channel URL state. Affected Packages / Versions - Package:...

Insertion of Sensitive Information Into Sent Data

Overview openclaw is a 🦞 OpenClaw — Personal AI Assistant Affected versions of this package are vulnerable to Insertion of Sensitive Information Into Sent Data via the OAuth onboarding process in the macOS beta application, where the PKCE codeverifier was exposed as the OAuth state in the URL. An...

TraceGuard: Process-Guided Firewall against Reasoning Backdoors in Large Language Models

The deployment of Large Reasoning Models LRMs in high-stakes decision-making pipelines has introduced a novel and opaque attack surface: reasoning backdoors. In these attacks, the model's intermediate Chain-of-Thought CoT is manipulated to provide a linguistically plausible but logically fallacio...

The vulnerability of the check_func_arg() function in the kernel/bpf/verifier.c module, which is part of the BPF interpreter support for the Linux operating system’s kernel, allows a attacker to compromise the confidentiality, integrity, and accessibility of the protected information.

The vulnerability of the checkfuncarg function in the kernel/bpf/verifier.c module, which is part of the BPF interpreter support for the Linux operating system’s kernel, relates to the lack of checking for the returned value. Exploiting this vulnerability could allow a remote attacker to compromi...

The vulnerability of the record_func_key() function in the kernel/bpf/verifier.c module, part of the BPF interpreter support for the Linux operating system’s kernel, allows a attacker to compromise the confidentiality, integrity, and accessibility of the protected information.

The vulnerability of the recordfunckey function in the kernel/bpf/verifier.c module, which is part of the BPF interpreter support in the Linux operating system’s kernel, involves reading beyond the buffer boundaries. Exploiting this vulnerability could allow an attacker to compromise the...

MalTool: Malicious Tool Attacks on LLM Agents

In a malicious tool attack, an attacker uploads a malicious tool to a distribution platform; once a user installs the tool and the LLM agent selects it during task execution, the tool can compromise the user's security and privacy. Prior work primarily focuses on manipulating tool names and...

K000159974: OpenSSL vulnerability CVE-2025-15469

Security Advisory Description Issue summary: The 'openssl dgst' command-line tool silently truncates input data to 16MB when using one-shot signing algorithms and reports success instead of an error. Impact summary: A user signing or verifying files larger than 16MB with one-shot algorithms such ...

The vulnerability of the regs_refineCond_op() function in the kernel/bpf/verifier.c module, which is part of the BPF interpreter support in the Linux operating system, allows a hacker to trigger a service failure.

The vulnerability of the regsrefineCondop function in the kernel/bpf/verifier.c module, which is part of the BPF interpreter in Linux operating systems, stems from the absence of a default block in the switch statement. Exploiting this vulnerability could allow an attacker to trigger a service...

Azure Linux 3.0 Security Update: kernel (CVE-2024-45020)

The version of kernel installed on the remote Azure Linux 3.0 host is prior to tested version. It is, therefore, affected by a vulnerability as referenced in the CVE-2024-45020 advisory. - In the Linux kernel, the following vulnerability has been resolved: bpf: Fix a kernel verifier crash in...

Triton VM has a Soundness Vulnerability due to Improper Sampling of Randomness

In affected versions of Triton VM, the verifier failed to correctly sample randomness in the FRI sub-protocol. Malicious provers can exploit this to craft proofs for arbitrary statements that this verifier accepts as valid, undermining soundness. Protocols that rely on proofs and the supplied...