GeoNode’s maintainer reports command execution through crafted vector-upload layer names. Its advisory lists versions through 5.0.1 as affected. GitHub’s reviewed database lists 5.1.0 as fixed, while the maintainer’s patched-version field remains unset.
Why it matters
The uploaded layer name reaches an operating-system command boundary. The source metadata disagrees on release coverage, so a database fixed-version entry alone should not establish a safe deployment.
Supported action
Restrict untrusted vector uploads while checking the upstream fix and release coverage. Confirm the patch is present in the deployed GeoNode build before restoring that upload path.
The Chainguard kaniko fork’s advisory describes Azure access-token exposure when untrusted Dockerfiles or build arguments reference lookalike registry hosts and an Azure identity is available. Its bundled credential helper can send a token to an attacker-controlled registry.
Why it mattersBuild inputs can cross the Azure identity trust boundary. Inventory both the executor and warmer, rather than assuming a separate credential-helper deployment is required.
OPSSHIELD says attackers used its cPGuard update channel to send a malicious update on November 2. Its later disclosure describes roughly 70 minutes of malicious distribution and exposure of root-level secrets, superseding the vendor’s earlier account.
Why it mattersA clean update cannot invalidate credentials copied before cleanup. The vendor’s earlier and later accounts describe different investigation stages; their scope estimates should not be combined.
MotherSphere’s NIKKE investigation describes tampered scripts on Level Infinite’s CDN loading a remote-control agent. Follow-up reporting describes credential-stealing capabilities and warns that updated launcher cleanup can erase evidence of an earlier infection.
Why it mattersCleaning the delivery scripts does not undo an earlier infection. The research draws on an infected machine and attributed user reports; it does not establish how many players were affected.
Rails’ November 4 advisory describes arbitrary file writes during attachment analysis when Active Storage uses ImageMagick, accepts untrusted uploads, and lacks a restrictive coder policy. Disk-backed storage can also expose files readable by the app.
Socket’s November 6 analysis links strutil-core 1.0.1, published November 1, to an install-time credential stealer and AI tool persistence. It was a standalone malicious gem; victims and successful downstream infections remain unconfirmed.
CVE-2026-98432 affects Linux IPv4 TCP connection handling. Aikido reports a demonstrated remote-root exploit against an MPTCP listener; that condition should guide urgent exposure checks without being treated as the full boundary of the flaw.
Cloudflare now reports certificates apparently forged with eMudhra’s weak intermediate keys and about 22,000 attempted handshakes in its telemetry. It has not seen interception of its own connections. Its challenge probes suggest some embedded browsers still accept affected chains, with a skewed traffic sample.
Cursor reported that tampered Claude Code extension builds were served from its marketplace on October 29. Microsoft found an injected loader that attempted SSH access and persistence. The evidence is specific to Cursor distribution; successful remote access depends on host conditions.
LiteLLM’s investigation connects an October 10 secret-stealing workflow to the October 21 publication of compromised litellm==1.107.0 artifacts. Socket independently found traces of the earlier workflow. The maintainer says token hashes link a stolen maintainer credential to approval of the later release.
Socket’s October investigation found malicious GitHub Actions workflows added through compromised maintainer accounts. Its initial case study covered 346 repositories; a later update identified a wider set of accounts and repositories. The workflow sought both CI secrets and credentials exposed in source or git history.
Socket’s package analysis identifies malicious releases in both the MemTensor npm plugin and the MemoryOS package on PyPI. The packages included a Go binary that searched developer environments for secrets. This is one distribution compromise spanning two registries, not two unrelated stories.
The PHP project released security updates for two engine flaws tracked as CVE-2026-109618 and CVE-2026-109619. Under the conditions described by the PHP Foundation, an unauthenticated request can reach code execution. This is a major runtime vulnerability, not a compromise of PHP packages or its release process.
The Node.js project rates CVE-2026-111497 high severity. Malformed Brotli data can drive an out-of-bounds write when an application decompresses attacker-supplied input through node:zlib. The affected supported release lines are 22.x, 24.x and 26.x. This is a runtime vulnerability, not evidence that Node.js distribution was compromised.
Evidence desk
Vulnerability Watch
Evidence snapshot ·
Recent CISA KEVs with EPSS probability at least 50% and percentile at least 95%. 9 qualified.
Apply mitigations in accordance with vendor instructions, ensuring compliance with CISA’s BOD 26-04 Prioritizing Security Updates Based on Risk (see URL in Notes) guidance and CISA’s “Forensics Triage Requirements” (see URL in Notes). Follow applicable BOD 26-04 guidance for cloud services or discontinue use of the product if mitigations are unavailable. Stakeholders are responsible for evaluating each asset's internet exposure and ensuring adherence to BOD 26-04 patching guidelines.
Apply mitigations in accordance with vendor instructions, ensuring compliance with CISA’s BOD 26-04 Prioritizing Security Updates Based on Risk (see URL in Notes) guidance and CISA’s “Forensics Triage Requirements” (see URL in Notes). Follow applicable BOD 26-04 guidance for cloud services or discontinue use of the product if mitigations are unavailable. Stakeholders are responsible for evaluating each asset's internet exposure and ensuring adherence to BOD 26-04 patching guidelines.
Apply mitigations in accordance with vendor instructions, ensuring compliance with CISA’s BOD 26-04 Prioritizing Security Updates Based on Risk (see URL in Notes) guidance and CISA’s “Forensics Triage Requirements” (see URL in Notes). Follow applicable BOD 26-04 guidance for cloud services or discontinue use of the product if mitigations are unavailable. Stakeholders are responsible for evaluating each asset's internet exposure and ensuring adherence to BOD 26-04 patching guidelines.
Apply mitigations in accordance with vendor instructions, ensuring compliance with CISA’s BOD 26-04 Prioritizing Security Updates Based on Risk (see URL in Notes) guidance and CISA’s “Forensics Triage Requirements” (see URL in Notes). Follow applicable BOD 26-04 guidance for cloud services or discontinue use of the product if mitigations are unavailable. Stakeholders are responsible for evaluating each asset's internet exposure and ensuring adherence to BOD 26-04 patching guidelines.
Apply mitigations in accordance with vendor instructions, ensuring compliance with CISA’s BOD 26-04 Prioritizing Security Updates Based on Risk (see URL in Notes) guidance and CISA’s “Forensics Triage Requirements” (see URL in Notes). Follow applicable BOD 26-04 guidance for cloud services or discontinue use of the product if mitigations are unavailable. Stakeholders are responsible for evaluating each asset's internet exposure and ensuring adherence to BOD 26-04 patching guidelines.
Apply mitigations in accordance with vendor instructions, ensuring compliance with CISA’s BOD 26-04 Prioritizing Security Updates Based on Risk (see URL in Notes) guidance and CISA’s “Forensics Triage Requirements” (see URL in Notes). Follow applicable BOD 26-04 guidance for cloud services or discontinue use of the product if mitigations are unavailable. Stakeholders are responsible for evaluating each asset's internet exposure and ensuring adherence to BOD 26-04 patching guidelines.
Apply mitigations in accordance with vendor instructions, ensuring compliance with CISA’s BOD 26-04 Prioritizing Security Updates Based on Risk (see URL in Notes) guidance and CISA’s “Forensics Triage Requirements” (see URL in Notes). Follow applicable BOD 26-04 guidance for cloud services or discontinue use of the product if mitigations are unavailable. Stakeholders are responsible for evaluating each asset's internet exposure and ensuring adherence to BOD 26-04 patching guidelines.
Collabora · Collabora Online – Built-in CODE Server
Apply mitigations in accordance with vendor instructions, ensuring compliance with CISA’s BOD 26-04 Prioritizing Security Updates Based on Risk (see URL in Notes) guidance and CISA’s “Forensics Triage Requirements” (see URL in Notes). Follow applicable BOD 26-04 guidance for cloud services or discontinue use of the product if mitigations are unavailable. Stakeholders are responsible for evaluating each asset's internet exposure and ensuring adherence to BOD 26-04 patching guidelines.
Apply mitigations in accordance with vendor instructions, ensuring compliance with CISA’s BOD 26-04 Prioritizing Security Updates Based on Risk (see URL in Notes) guidance and CISA’s “Forensics Triage Requirements” (see URL in Notes). Follow applicable BOD 26-04 guidance for cloud services or discontinue use of the product if mitigations are unavailable. Stakeholders are responsible for evaluating each asset's internet exposure and ensuring adherence to BOD 26-04 patching guidelines.
GeoNode’s maintainer reports command execution through crafted vector-upload layer names. Its advisory lists versions through 5.0.1 as affected. GitHub’s reviewed database lists 5.1.0 as fixed, while the maintainer’s patched-version field remains unset.
The Chainguard kaniko fork’s advisory describes Azure access-token exposure when untrusted Dockerfiles or build arguments reference lookalike registry hosts and an Azure identity is available. Its bundled credential helper can send a token to an attacker-controlled registry.
OPSSHIELD says attackers used its cPGuard update channel to send a malicious update on November 2. Its later disclosure describes roughly 70 minutes of malicious distribution and exposure of root-level secrets, superseding the vendor’s earlier account.
MotherSphere’s NIKKE investigation describes tampered scripts on Level Infinite’s CDN loading a remote-control agent. Follow-up reporting describes credential-stealing capabilities and warns that updated launcher cleanup can erase evidence of an earlier infection.
Rails’ November 4 advisory describes arbitrary file writes during attachment analysis when Active Storage uses ImageMagick, accepts untrusted uploads, and lacks a restrictive coder policy. Disk-backed storage can also expose files readable by the app.
Socket’s November 6 analysis links strutil-core 1.0.1, published November 1, to an install-time credential stealer and AI tool persistence. It was a standalone malicious gem; victims and successful downstream infections remain unconfirmed.
CVE-2026-98432 affects Linux IPv4 TCP connection handling. Aikido reports a demonstrated remote-root exploit against an MPTCP listener; that condition should guide urgent exposure checks without being treated as the full boundary of the flaw.
Cloudflare now reports certificates apparently forged with eMudhra’s weak intermediate keys and about 22,000 attempted handshakes in its telemetry. It has not seen interception of its own connections. Its challenge probes suggest some embedded browsers still accept affected chains, with a skewed traffic sample.
Cursor reported that tampered Claude Code extension builds were served from its marketplace on October 29. Microsoft found an injected loader that attempted SSH access and persistence. The evidence is specific to Cursor distribution; successful remote access depends on host conditions.
LiteLLM’s investigation connects an October 10 secret-stealing workflow to the October 21 publication of compromised litellm==1.107.0 artifacts. Socket independently found traces of the earlier workflow. The maintainer says token hashes link a stolen maintainer credential to approval of the later release.
Socket’s October investigation found malicious GitHub Actions workflows added through compromised maintainer accounts. Its initial case study covered 346 repositories; a later update identified a wider set of accounts and repositories. The workflow sought both CI secrets and credentials exposed in source or git history.
Socket’s package analysis identifies malicious releases in both the MemTensor npm plugin and the MemoryOS package on PyPI. The packages included a Go binary that searched developer environments for secrets. This is one distribution compromise spanning two registries, not two unrelated stories.
The PHP project released security updates for two engine flaws tracked as CVE-2026-109618 and CVE-2026-109619. Under the conditions described by the PHP Foundation, an unauthenticated request can reach code execution. This is a major runtime vulnerability, not a compromise of PHP packages or its release process.
The Node.js project rates CVE-2026-111497 high severity. Malformed Brotli data can drive an out-of-bounds write when an application decompresses attacker-supplied input through node:zlib. The affected supported release lines are 22.x, 24.x and 26.x. This is a runtime vulnerability, not evidence that Node.js distribution was compromised.