Real vulnerabilities explained in plain language, with sources you can check: who is affected, how the attack works, and how to fix and detect it.https://plaincve.date/enhttps://plaincve.date/en/vulns/cve-2026-12537-gemini-cli-headless-workspace-trust/https://plaincve.date/en/vulns/cve-2026-12537-gemini-cli-headless-workspace-trust/In CI, Gemini CLI trusted any workspace automatically, so a malicious .gemini/.env in a pull request could run commands on the build host.Thu, 24 Sep 2026 00:00:00 GMTgemini-cligithub-actionsci-cdworkspace-trustai-agentcommand-injectionCVE-2026-12537https://plaincve.date/en/vulns/cve-2026-33017-langflow-public-flow-rce/https://plaincve.date/en/vulns/cve-2026-33017-langflow-public-flow-rce/Langflow's public flow endpoint ran Python code sent by anyone, with no login, letting attackers take over the server. Exploited in the wild.Thu, 24 Sep 2026 00:00:00 GMTlangflowrcecode-injectionmissing-authenticationpythonai-pipelineCVE-2026-33017https://plaincve.date/en/vulns/cve-2026-45758-guardrails-ai-pypi-compromise/https://plaincve.date/en/vulns/cve-2026-45758-guardrails-ai-pypi-compromise/A malicious guardrails-ai 0.10.1 was pushed to PyPI with a stolen token; on Linux, importing it fetched and ran a remote payload.Thu, 24 Sep 2026 00:00:00 GMTpypipythonmalicious-packageci-cdcredential-theftshai-huludllm-guardrailsCVE-2026-45758https://plaincve.date/en/vulns/cve-2026-48501-github-cli-token-leak/https://plaincve.date/en/vulns/cve-2026-48501-github-cli-token-leak/GitHub CLI before 2.93.0 could attach your GitHub login token to requests to TUF and artifact hosts during three verification commands.Thu, 24 Sep 2026 00:00:00 GMTgithub-clicredential-leaktokensigstoretufattestationgoCVE-2026-48501https://plaincve.date/en/vulns/cve-2026-50016-pnpm-alias-path-traversal/https://plaincve.date/en/vulns/cve-2026-50016-pnpm-alias-path-traversal/pnpm didn't block '..' in dependency aliases, so a malicious package deep in the tree can swap project paths for symlinks, even with --ignore-scripts.Thu, 24 Sep 2026 00:00:00 GMTpnpmnpmjavascriptpath-traversalsymlinkignore-scriptstransitive-dependencyCVE-2026-50016https://plaincve.date/en/vulns/cve-2026-5241-transformers-lightglue-remote-code-override/https://plaincve.date/en/vulns/cve-2026-5241-transformers-lightglue-remote-code-override/In Transformers before 5.5.0, a LightGlue model's own config could override trust_remote_code=False and run the model's bundled Python code.Thu, 24 Sep 2026 00:00:00 GMThuggingfacetransformerstrust-remote-codemodel-supply-chainpythoncode-executionCVE-2026-5241https://plaincve.date/en/vulns/cve-2026-82021-hermes-agent-mcp-unpinned-branch/https://plaincve.date/en/vulns/cve-2026-82021-hermes-agent-mcp-unpinned-branch/Hermes Agent's MCP catalog installed its n8n bridge from a movable branch; if that upstream repo were hijacked, installs would run its code.Thu, 24 Sep 2026 00:00:00 GMTmcpai-agentversion-pinninggitn8nintegrity-checkCVE-2026-82021https://plaincve.date/en/vulns/cve-2026-90553-vllm-llavaonevision2-remote-code-ignored/https://plaincve.date/en/vulns/cve-2026-90553-vllm-llavaonevision2-remote-code-ignored/vLLM before 0.28.0 ignores trust_remote_code=False for LlavaOnevision2 models, so code bundled in a malicious model can still run.Thu, 24 Sep 2026 00:00:00 GMTvllmtrust-remote-codemodel-supply-chainpythonkwargsmultimodalCVE-2026-90553https://plaincve.date/en/vulns/cve-2026-26030-semantic-kernel-filter-rce/https://plaincve.date/en/vulns/cve-2026-26030-semantic-kernel-filter-rce/Semantic Kernel turns the filter an AI model writes for a search plugin into Python code and runs it; prompt injection can make that run any code on the server.Wed, 23 Sep 2026 00:00:00 GMTprompt-injectionllmagentpythonevalrceCVE-2026-26030https://plaincve.date/en/vulns/cve-2026-33634-trivy-supply-chain/https://plaincve.date/en/vulns/cve-2026-33634-trivy-supply-chain/Attackers hijacked Trivy's releases to ship a credential stealer; secrets taken from CI pipelines that ran it were then used to poison LiteLLM and others.Wed, 23 Sep 2026 00:00:00 GMTgithub-actionsci-cdcredential-theftpypidocker-hubtrivylitellmCVE-2026-33634https://plaincve.date/en/vulns/cve-2026-45321-tanstack-npm-supply-chain-compromise/https://plaincve.date/en/vulns/cve-2026-45321-tanstack-npm-supply-chain-compromise/Attackers hijacked TanStack's GitHub release pipeline and pushed credential-stealing malware in 84 versions of 42 @tanstack/* npm packages.Wed, 23 Sep 2026 00:00:00 GMTnpmgithub-actionspull-request-targetcache-poisoningoidcwormmini-shai-huludcredential-theftkevCVE-2026-45321https://plaincve.date/en/vulns/cve-2026-59822-litellm-mcp-auth-bypass/https://plaincve.date/en/vulns/cve-2026-59822-litellm-mcp-auth-bypass/LiteLLM Proxy's MCP endpoints let failed logins through as anonymous, so anyone can list and call its MCP tools with no account. Exploited in the wild.Wed, 23 Sep 2026 00:00:00 GMTmcpllm-gatewayauthenticationpythonfail-openCVE-2026-59822https://plaincve.date/en/vulns/cve-2026-94127-f5-big-ip-apm-oauth-heap-overflow/https://plaincve.date/en/vulns/cve-2026-94127-f5-big-ip-apm-oauth-heap-overflow/An F5 BIG-IP APM memory bug lets attackers who have not logged in run code on systems acting as OAuth Authorization Servers. Exploited as a zero-day.Wed, 23 Sep 2026 00:00:00 GMTf5big-ipapmoauthheap-overflowrcezero-daykevedge-deviceCVE-2026-94127https://plaincve.date/en/vulns/ghsa-2026-axios-npm-compromise/https://plaincve.date/en/vulns/ghsa-2026-axios-npm-compromise/Attackers hijacked an axios maintainer's npm account and published two versions with a malicious dependency; a plain npm install drops a remote access trojan.Wed, 23 Sep 2026 00:00:00 GMTnpmjavascriptaccount-takeoverpostinstallratnorth-korea