https://vulnpocalypse.dev/feeds/Advisories for AI coding assistants, agents, MCP servers and AI platforms: prompt injection, leaked keys and poisoned tools.enThu, 12 Nov 2026 16:59:02 +0000https://vulnpocalypse.dev/advisories/vibe-coded-apis-ex-hacker-says-they-are-riddled-with-flaws/https://vulnpocalypse.dev/advisories/vibe-coded-apis-ex-hacker-says-they-are-riddled-with-flaws/Thu, 12 Nov 2026 12:00:00 +0000Bank Info Security reports an ex-hacker says vibe-coded APIs are riddled with flaws; details unconfirmed, so review AI-generated API code.ai-toolsai-toolswebhttps://vulnpocalypse.dev/advisories/darktrace-says-stolen-claude-api-keys-can-steer-claude-code-as-implant/https://vulnpocalypse.dev/advisories/darktrace-says-stolen-claude-api-keys-can-steer-claude-code-as-implant/Thu, 12 Nov 2026 12:00:00 +0000Help Net Security reports Darktrace showed stolen Claude API keys can steer Claude Code in a lab test; scope and rotate keys.ai-toolsai-toolsendpointhttps://vulnpocalypse.dev/advisories/cyble-says-agentic-ai-can-investigate-phishing-with-human-sign-off/https://vulnpocalypse.dev/advisories/cyble-says-agentic-ai-can-investigate-phishing-with-human-sign-off/Thu, 12 Nov 2026 12:00:00 +0000Cyble says agentic AI can run most of a phishing investigation but warns of hidden prompts in emails; keep a human approving verdicts.ai-toolsai-toolshttps://vulnpocalypse.dev/advisories/akamai-says-ai-found-bugs-mean-rethinking-exposure-response/https://vulnpocalypse.dev/advisories/akamai-says-ai-found-bugs-mean-rethinking-exposure-response/Thu, 12 Nov 2026 12:00:00 +0000Akamai published a blog post on AI finding vulnerabilities and rethinking exposure; only the headline was reviewed, so no action is confirmed.ai-toolsai-toolsvuln-datahttps://vulnpocalypse.dev/advisories/chatgpt-shared-link-mathlive-xss/https://vulnpocalypse.dev/advisories/chatgpt-shared-link-mathlive-xss/Thu, 12 Nov 2026 12:00:00 +0000A public ChatGPT share link could run attacker JavaScript as chatgpt.com and read private chats. OpenAI fixed it on 26 Oct; MathLive users should update.ai-toolsai-toolswebnpmhttps://vulnpocalypse.dev/advisories/cursor-2026-cves/https://vulnpocalypse.dev/advisories/cursor-2026-cves/Thu, 12 Nov 2026 12:00:00 +0000Seven 2026 Cursor CVEs let prompt injection or a cloned repo run code outside the sandbox. Update to 3.1.2 or later; the Windows git.exe bug has no stated fix.ai-toolsai-toolsendpointhttps://vulnpocalypse.dev/advisories/lazarus-microsoft-shows-hidden-backdoor-in-ai-agent-model-weights/https://vulnpocalypse.dev/advisories/lazarus-microsoft-shows-hidden-backdoor-in-ai-agent-model-weights/Wed, 11 Nov 2026 12:00:00 +0000Microsoft researchers showed that editing fewer than 10 neurons can backdoor an AI agent model; verify model provenance and limit agent permissions.supply-chainai-toolshttps://vulnpocalypse.dev/advisories/grok-app-builder-35578-public-grokme-apps-expose-data-and-passwords/https://vulnpocalypse.dev/advisories/grok-app-builder-35578-public-grokme-apps-expose-data-and-passwords/Wed, 11 Nov 2026 12:00:00 +0000Researchers found 35,578 of 37,571 Grok-built apps open without sign-in, some with passwords; check app access settings and unpublish unowned apps.ai-toolsai-toolswebcloudhttps://vulnpocalypse.dev/advisories/endor-labs-reports-openclaw-apply-patch-path-traversal-fixed-in-2026214/https://vulnpocalypse.dev/advisories/endor-labs-reports-openclaw-apply-patch-path-traversal-fixed-in-2026214/Wed, 11 Nov 2026 12:00:00 +0000Endor Labs reports a path traversal in OpenClaw apply_patch; the project advisory lists a fix in 2026.2.14, so upgrade openclaw.ai-toolsnpmnodejsai-toolsopen-sourcehttps://vulnpocalypse.dev/advisories/endor-labs-claims-its-ai-sast-found-26x-more-real-flaws-than-claude/https://vulnpocalypse.dev/advisories/endor-labs-claims-its-ai-sast-found-26x-more-real-flaws-than-claude/Wed, 11 Nov 2026 12:00:00 +0000Endor Labs claims its AI SAST found 192 real vulnerabilities in an 8-project benchmark; treat it as a vendor claim and test on your own code.ai-toolsai-toolsopen-sourcehttps://vulnpocalypse.dev/advisories/databreachesnet-reports-grok-security-gap-exposed-users-and-musk-secrets/https://vulnpocalypse.dev/advisories/databreachesnet-reports-grok-security-gap-exposed-users-and-musk-secrets/Wed, 11 Nov 2026 12:00:00 +0000DataBreaches.net reports a Grok security blind spot exposed users and Musk company secrets; details are not confirmed, so check official statements.ai-toolsai-toolshttps://vulnpocalypse.dev/advisories/cloudsek-reports-exposed-silent-lynx-server-tied-to-central-asia-espionage/https://vulnpocalypse.dev/advisories/cloudsek-reports-exposed-silent-lynx-server-tied-to-central-asia-espionage/Wed, 11 Nov 2026 12:00:00 +0000CloudSEK reports an exposed Silent Lynx server tied to Central Asia espionage; defenders there can check logs for the listed IP address.ai-toolsenterpriseai-toolshttps://vulnpocalypse.dev/advisories/australia-to-mandate-ai-breach-reporting-after-openai-agent-hit-medicare-site/https://vulnpocalypse.dev/advisories/australia-to-mandate-ai-breach-reporting-after-openai-agent-hit-medicare-site/Wed, 11 Nov 2026 12:00:00 +0000Australia will make AI firms report breaches after an OpenAI agent accessed a Medicare portal; AI operators should review incident reporting.ai-toolsai-toolswebhttps://vulnpocalypse.dev/advisories/operation-snowball-jailbroken-ai-agent-hits-european-telecom/https://vulnpocalypse.dev/advisories/operation-snowball-jailbroken-ai-agent-hits-european-telecom/Wed, 11 Nov 2026 12:00:00 +0000A jailbroken AI agent attacked a European telecom and 13 hosted organizations for two weeks; patch Cisco BroadWorks CVE-2026-20320 now.ai-toolsai-toolsnetwork-edgeenterprisecloudhttps://vulnpocalypse.dev/advisories/claude-code-settingslocaljson-leaks-secrets-to-public-github-repos/https://vulnpocalypse.dev/advisories/claude-code-settingslocaljson-leaks-secrets-to-public-github-repos/Wed, 11 Nov 2026 12:00:00 +0000Claude Code can save approved commands with secrets in .claude/settings.local.json, found in public GitHub repos; rotate any credential it ever held.ai-toolsai-toolsopen-sourcehttps://vulnpocalypse.dev/advisories/claude-dispatch-unpatched-rce/https://vulnpocalypse.dev/advisories/claude-dispatch-unpatched-rce/Wed, 11 Nov 2026 12:00:00 +0000A browser extension on claude.ai can drive Claude Dispatch to run commands on connected desktops. Anthropic closed it as a duplicate and won't fix it.ai-toolsai-toolsendpointhttps://vulnpocalypse.dev/advisories/openai-gpt6-cyber-948-linux-flaws/https://vulnpocalypse.dev/advisories/openai-gpt6-cyber-948-linux-flaws/Wed, 11 Nov 2026 12:00:00 +0000OpenAI's GPT-6 Cyber built 948 working Linux kernel root exploits and reported 81 high-severity Firefox flaws, all disclosed for patching.ai-toolsai-toolsopen-sourceendpointhttps://vulnpocalypse.dev/advisories/artex-ai-framework-exposed-korean-banks/https://vulnpocalypse.dev/advisories/artex-ai-framework-exposed-korean-banks/Wed, 11 Nov 2026 12:00:00 +0000Operators of the China-made ARTEX AI pentest framework left servers open, exposing Claude Code and DeepSeek-driven tooling tied to breaches at seven South.ai-toolsai-toolsenterprisecloudopen-sourcehttps://vulnpocalypse.dev/advisories/workos-artex-ai-breach/https://vulnpocalypse.dev/advisories/workos-artex-ai-breach/Wed, 11 Nov 2026 12:00:00 +0000On 30 October 2026 WorkOS found an attacker had used the ARTEX AI pentest agent to breach an internal sales app and read CRM and customer records.ai-toolsai-toolsenterprisecloudhttps://vulnpocalypse.dev/advisories/ncsc-warns-ai-powered-ransomware-is-highly-likely-to-hit-uk-critical-infrastruct/https://vulnpocalypse.dev/advisories/ncsc-warns-ai-powered-ransomware-is-highly-likely-to-hit-uk-critical-infrastruct/Wed, 11 Nov 2026 12:00:00 +0000The UK NCSC says AI-powered malware is highly likely to feature in ransomware attacks on critical infrastructure by 2028, so patch and fix the basics now.ai-toolsai-toolsenterprisephphttps://vulnpocalypse.dev/advisories/llmjacking-google-researchers-report-stolen-ai-access-bills-up-to-1-million/https://vulnpocalypse.dev/advisories/llmjacking-google-researchers-report-stolen-ai-access-bills-up-to-1-million/Wed, 11 Nov 2026 12:00:00 +0000Google researchers report LLMJacking, where stolen cloud credentials run AI models at victim cost; watch AI service changes and rotate keys.ai-toolsai-toolscloudenterprisehttps://vulnpocalypse.dev/advisories/3am-ransomware-encrypts-37-machines-in-uk-school-estate-in-suspected-ai-assisted/https://vulnpocalypse.dev/advisories/3am-ransomware-encrypts-37-machines-in-uk-school-estate-in-suspected-ai-assisted/Wed, 11 Nov 2026 12:00:00 +0000A 3AM ransomware affiliate encrypted 37 machines in a UK school estate in September 2026, with Bitdefender suspecting AI-assisted command retries.ai-toolsenterpriseendpointai-toolshttps://vulnpocalypse.dev/advisories/hugging-face-uploader-claims-leaked-anthropic-numbat-model-checkpoint/https://vulnpocalypse.dev/advisories/hugging-face-uploader-claims-leaked-anthropic-numbat-model-checkpoint/Wed, 11 Nov 2026 12:00:00 +0000An anonymous Hugging Face uploader claims to share a leaked Anthropic training checkpoint; do not run its bundled custom code outside isolation.ai-toolsai-toolshttps://vulnpocalypse.dev/advisories/volta-ai-infostealer/https://vulnpocalypse.dev/advisories/volta-ai-infostealer/Wed, 11 Nov 2026 12:00:00 +0000Cisco Talos found VOLTA, a Windows infostealer that grabs credentials, crypto wallets and Gmail data, then asks Claude, Kimi or Gemini to rank the victim.ai-toolsendpointai-toolshttps://vulnpocalypse.dev/advisories/northflank-breach-ai-key-abuse/https://vulnpocalypse.dev/advisories/northflank-breach-ai-key-abuse/Wed, 11 Nov 2026 12:00:00 +0000Northflank disclosed a breach on 30 Oct 2026 after Wiz flagged it; customers report unauthorised AI API key use. Treat all env vars as compromised and rotate.ai-toolscloudai-toolshttps://vulnpocalypse.dev/advisories/solanacry-fake-solana-github-repo-lures-ai-coding-agents-hackread-reports/https://vulnpocalypse.dev/advisories/solanacry-fake-solana-github-repo-lures-ai-coding-agents-hackread-reports/Wed, 11 Nov 2026 12:00:00 +0000Hackread reports a fake GitHub Solana starter kit tells AI agents to install NetSupport RAT; review commands before agents run them.supply-chainai-toolsopen-sourcecryptoendpointhttps://vulnpocalypse.dev/advisories/chainalysis-breach-by-offensive-agentic-techniques-exposes-customer-names-and-em/https://vulnpocalypse.dev/advisories/chainalysis-breach-by-offensive-agentic-techniques-exposes-customer-names-and-em/Wed, 11 Nov 2026 12:00:00 +0000Chainalysis says an attacker using offensive agentic techniques accessed customer user IDs, names, emails and role labels, but not investigation content orai-toolsai-toolscryptoenterprisehttps://vulnpocalypse.dev/advisories/black-october-ai-platform-breaches/https://vulnpocalypse.dev/advisories/black-october-ai-platform-breaches/Wed, 11 Nov 2026 12:00:00 +0000Nine AI developer platforms disclosed breaches in October 2026. Stolen keys from stored secrets drove about 2.5M Bedrock calls. Revoke, then rotate.ai-toolscloudai-toolshttps://vulnpocalypse.dev/advisories/trojanized-claude-code-extension-21312-21313-on-cursor-marketplace-installs-pers/https://vulnpocalypse.dev/advisories/trojanized-claude-code-extension-21312-21313-on-cursor-marketplace-installs-pers/Wed, 11 Nov 2026 12:00:00 +0000A tampered anthropic.claude-code build on the Cursor marketplace planted an SSH key, a reverse tunnel and persistence on up to 5,000 developer machines.supply-chainai-toolsendpointhttps://vulnpocalypse.dev/advisories/sophos-reports-gold-ogre-used-ai-agents-in-aur0ra-ransomware-intrusions/https://vulnpocalypse.dev/advisories/sophos-reports-gold-ogre-used-ai-agents-in-aur0ra-ransomware-intrusions/Wed, 11 Nov 2026 12:00:00 +0000Sophos reports the GOLD OGRE group used Cursor and AI models to run Aur0ra ransomware intrusions, so check for its indicators and tighten AD defenses.ai-toolsai-toolsenterpriseendpointhttps://vulnpocalypse.dev/advisories/cursor-claude-code-extension-backdoor/https://vulnpocalypse.dev/advisories/cursor-claude-code-extension-backdoor/Wed, 11 Nov 2026 12:00:00 +0000Fake Claude Code 2.1.312 and 2.1.313 on Cursor's marketplace added an attacker SSH key and reverse tunnel. Check authorized_keys for svc-telemetry.supply-chainai-toolsendpointopen-sourcehttps://vulnpocalypse.dev/advisories/claude-code-extension-cursor-trojanized/https://vulnpocalypse.dev/advisories/claude-code-extension-cursor-trojanized/Wed, 11 Nov 2026 12:00:00 +0000A tampered Anthropic.claude-code extension (2.1.312/2.1.313) was live on Cursor's marketplace ~11 hours, planting an SSH backdoor. Rotate keys, remove it.supply-chainai-toolsendpointhttps://vulnpocalypse.dev/advisories/google-gemini-4-carbon-security/https://vulnpocalypse.dev/advisories/google-gemini-4-carbon-security/Wed, 11 Nov 2026 12:00:00 +0000Google released Gemini 4 Carbon (28 Oct 2026), a model it says leads at finding software vulnerabilities, via its limited Fairwind program. Not a breach.ai-toolsai-toolscloudhttps://vulnpocalypse.dev/advisories/anthropic-ai-model-chained-flaws-into-chrome-safari-android-windows-and-linux-ex/https://vulnpocalypse.dev/advisories/anthropic-ai-model-chained-flaws-into-chrome-safari-android-windows-and-linux-ex/Wed, 11 Nov 2026 12:00:00 +0000Anthropic says an unreleased model chained flaws into five working exploits for Chrome, Safari, Android, Windows and Linux, all reported and patched.ai-toolsai-toolswebendpointopen-sourcehttps://vulnpocalypse.dev/advisories/xint-says-openai-api-compaction-flaw-exposed-hidden-model-reasoning/https://vulnpocalypse.dev/advisories/xint-says-openai-api-compaction-flaw-exposed-hidden-model-reasoning/Wed, 11 Nov 2026 12:00:00 +0000Xint reports an OpenAI API compaction flaw exposed raw model reasoning; per Xint, OpenAI deployed a fix, and no client action is described.ai-toolsai-toolscloudhttps://vulnpocalypse.dev/advisories/symantec-reports-ai-attack-agent-built-to-hijack-exposed-model-endpoints/https://vulnpocalypse.dev/advisories/symantec-reports-ai-attack-agent-built-to-hijack-exposed-model-endpoints/Wed, 11 Nov 2026 12:00:00 +0000Symantec reports an AI attack agent built to hijack unauthenticated OpenAI-style endpoints, so require authentication on any model endpoint you expose.ai-toolsai-toolswebnextjscloudcryptohttps://vulnpocalypse.dev/advisories/railway-breach-rotate-all-environment-variables-and-secrets/https://vulnpocalypse.dev/advisories/railway-breach-rotate-all-environment-variables-and-secrets/Wed, 11 Nov 2026 12:00:00 +0000Railway confirmed attackers may have decrypted customer environment variables; rotate every secret stored there and review provider logs.ai-toolscloudci-cdai-toolshttps://vulnpocalypse.dev/advisories/litellm-pypi-docker-compromise/https://vulnpocalypse.dev/advisories/litellm-pypi-docker-compromise/Wed, 11 Nov 2026 12:00:00 +0000Attackers with stolen gh CLI tokens shipped malware as LiteLLM 1.107.0 on PyPI, Docker Hub and GHCR. Check before running Python, then rebuild and rotate.supply-chainpypiai-toolsci-cdhttps://vulnpocalypse.dev/advisories/litellm-11070-on-pypi-and-docker-was-backdoored-with-a-credential-stealer-and-re/https://vulnpocalypse.dev/advisories/litellm-11070-on-pypi-and-docker-was-backdoored-with-a-credential-stealer-and-re/Wed, 11 Nov 2026 12:00:00 +0000LiteLLM 1.107.0 on PyPI and official Docker images ran a credential stealer on every Python start; roll back to 1.106.2 and rotate all reachable secrets.supply-chainpypiai-toolsopen-sourceci-cdcloudhttps://vulnpocalypse.dev/advisories/ovhcloud-cli-release-assets-infostealer/https://vulnpocalypse.dev/advisories/ovhcloud-cli-release-assets-infostealer/Wed, 11 Nov 2026 12:00:00 +0000OVHcloud CLI binaries for v0.11.0 to v0.15.0 fetched on 19 Oct 2026, 12:47 to 14:21 UTC, were an infostealer. Check MD5s, rotate secrets, install v0.16.0.supply-chaincloudendpointai-toolshttps://vulnpocalypse.dev/advisories/opencode-console-provider-key-breach/https://vulnpocalypse.dev/advisories/opencode-console-provider-key-breach/Wed, 11 Nov 2026 12:00:00 +0000An intruder reached an OpenCode legacy system and may have exposed provider keys saved in OpenCode Console. Revoke those keys at OpenAI, Anthropic or Google.ai-toolsai-toolscloudhttps://vulnpocalypse.dev/advisories/crafty-falcon-ai-attacks-indian-nbfcs-payment-apis/https://vulnpocalypse.dev/advisories/crafty-falcon-ai-attacks-indian-nbfcs-payment-apis/Wed, 11 Nov 2026 12:00:00 +0000CERT-In advisory CIAD-2026-0058 names CRAFTY FALCON, an AI-assisted actor invoking Indian NBFC and fintech payment APIs to move funds at machine speed.ai-toolsai-toolsenterprisecloudwebcryptohttps://vulnpocalypse.dev/advisories/artex-ai-pentest-framework-abused-in-data-theft-from-south-korean-banks/https://vulnpocalypse.dev/advisories/artex-ai-pentest-framework-abused-in-data-theft-from-south-korean-banks/Wed, 11 Nov 2026 12:00:00 +0000A suspected Chinese-speaking, financially motivated actor used the ARTEX agentic pentest tool to steal data from South Korean financial firms.ai-toolsai-toolsopen-sourcewebenterprisehttps://vulnpocalypse.dev/advisories/actions-cool-issues-helper-reenabled-malware/https://vulnpocalypse.dev/advisories/actions-cool-issues-helper-reenabled-malware/Wed, 11 Nov 2026 12:00:00 +0000Hijacked actions-cool/issues-helper and maintain-one-comment were re-enabled on 16 Sep 2026 with every tag still malicious. Remove them and rotate CI secrets.supply-chaingithub-actionsci-cdai-toolshttps://vulnpocalypse.dev/advisories/ai-coding-agents-ci-prompt-injection/https://vulnpocalypse.dev/advisories/ai-coding-agents-ci-prompt-injection/Wed, 11 Nov 2026 12:00:00 +0000Prompt injection in GitHub issues let attackers reach CI secrets through Claude Code, Gemini CLI and Codex. Upgrade, and keep untrusted input away from secrets.ai-toolsgithub-actionsai-toolsci-cdhttps://vulnpocalypse.dev/advisories/ai-found-vulnerabilities-exploitation-rate/https://vulnpocalypse.dev/advisories/ai-found-vulnerabilities-exploitation-rate/Wed, 11 Nov 2026 12:00:00 +0000VulnCheck found 1.3% of AI-found vulnerabilities exploited in 1H 2026, the same as all bugs. Patch AI-found CVEs by exposure, not by how they were found.ai-toolsai-toolsvuln-dataopen-sourcehttps://vulnpocalypse.dev/advisories/teampcp-trivy-litellm-supply-chain/https://vulnpocalypse.dev/advisories/teampcp-trivy-litellm-supply-chain/Wed, 11 Nov 2026 12:00:00 +0000Poisoned Trivy and KICS actions and images, LiteLLM 1.82.7/1.82.8 and Telnyx 4.87.1/4.87.2 stole CI and developer secrets. Check, then rotate.supply-chaingithub-actionspypici-cdai-toolshttps://vulnpocalypse.dev/advisories/vulncheck-says-anthropic-glasswing-cves-are-few-fixes-lag/https://vulnpocalypse.dev/advisories/vulncheck-says-anthropic-glasswing-cves-are-few-fixes-lag/Tue, 10 Nov 2026 12:00:00 +0000VulnCheck says 442 CVEs trace to Anthropic's Glasswing, 2 exploited; check those two CVEs and treat scanner reports as unverified.ai-toolsai-toolsopen-sourcevuln-datahttps://vulnpocalypse.dev/advisories/thai-nsc-says-cyberattacks-occur-daily-and-attackers-use-ai-per-report/https://vulnpocalypse.dev/advisories/thai-nsc-says-cyberattacks-occur-daily-and-attackers-use-ai-per-report/Tue, 10 Nov 2026 12:00:00 +0000Asia News Network reports Thai NSC says cyberattack attempts occur daily and attackers use AI; no specific flaw is named, so watch official guidance.ai-toolsai-toolsenterprisehttps://vulnpocalypse.dev/advisories/reversinglabs-reports-ai-linked-malware-signals-rising-no-clear-attack-surge/https://vulnpocalypse.dev/advisories/reversinglabs-reports-ai-linked-malware-signals-rising-no-clear-attack-surge/Tue, 10 Nov 2026 12:00:00 +0000ReversingLabs says LLM API references in malicious files are rising but no AI attack surge is shown; review exposure to PHP2Shell CVEs.ai-toolsai-toolsphpvuln-data