# What your agent missed in March 2026 In March 2026, CISA added 5 actively exploited vulnerabilities in web-stack software to its Known Exploited Vulnerabilities catalog, GitHub published 13 critical advisories for widely used npm and PyPI packages, 2 release lines reached end of life and 1 new release line shipped. The ones to know first: - **Mar 26, 2026**: CVE-2026-33634 added to CISA’s exploited list (Trivy). Aquasecurity Trivy contains an embedded malicious code vulnerability that could allow an attacker to gain access to everything in the CI/CD environment, including all… [CISA KEV](https://www.cisa.gov/known-exploited-vulnerabilities-catalog?field_cve=CVE-2026-33634) - **Mar 31, 2026**: fastmcp: SSRF & path traversal vulnerability (critical). Fixed in 3.2.0. [GitHub Advisory Database](https://github.com/advisories/GHSA-vv7q-7jx5-f767) - **Mar 31, 2026**: Ruby 3.2 reached end of life. No more security fixes. Latest: Ruby 4.0. [endoflife.date](https://endoflife.date/ruby) ## Exploited in the wild (5) - **Mar 26, 2026**: CVE-2026-33634 added to CISA’s exploited list (Trivy). Aquasecurity Trivy contains an embedded malicious code vulnerability that could allow an attacker to gain access to everything in the CI/CD environment, including all… [CISA KEV](https://www.cisa.gov/known-exploited-vulnerabilities-catalog?field_cve=CVE-2026-33634) - **Mar 25, 2026**: CVE-2026-33017 added to CISA’s exploited list (Langflow). Langflow contains a code injection vulnerability that could allow building public flows without requiring authentication. [CISA KEV](https://www.cisa.gov/known-exploited-vulnerabilities-catalog?field_cve=CVE-2026-33017) - **Mar 20, 2026**: CVE-2025-32432 added to CISA’s exploited list (Craft CMS). Craft CMS contains a code injection vulnerability that allows a remote attacker to execute arbitrary code. [CISA KEV](https://www.cisa.gov/known-exploited-vulnerabilities-catalog?field_cve=CVE-2025-32432) - **Mar 20, 2026**: CVE-2025-54068 added to CISA’s exploited list (Laravel Livewire). Laravel Livewire contain a code injection vulnerability that could allow unauthenticated attackers to achieve remote command execution in specific scenarios. [CISA KEV](https://www.cisa.gov/known-exploited-vulnerabilities-catalog?field_cve=CVE-2025-54068) - **Mar 11, 2026**: CVE-2025-68613 added to CISA’s exploited list (n8n). n8n contains an improper control of dynamically managed code resources vulnerability in its workflow expression evaluation system that allows for remote code execution. [CISA KEV](https://www.cisa.gov/known-exploited-vulnerabilities-catalog?field_cve=CVE-2025-68613) ## Critical advisories (13) - **Mar 31, 2026**: fastmcp: SSRF & path traversal vulnerability (critical). Fixed in 3.2.0. [GitHub Advisory Database](https://github.com/advisories/GHSA-vv7q-7jx5-f767) - **Mar 31, 2026**: mlflow: command injection when serving models with enable\_mlserver=True (critical). Fixed in 3.9.0. [GitHub Advisory Database](https://github.com/advisories/GHSA-rvhj-8chj-8v3c) - **Mar 30, 2026**: mlflow: MLflow command injection vulnerability (critical). Fixed in 3.8.1. [GitHub Advisory Database](https://github.com/advisories/GHSA-r23q-823p-vmf7) - **Mar 30, 2026**: mlflow: MLFlow path traversal vulnerability (critical). [GitHub Advisory Database](https://github.com/advisories/GHSA-vhcx-3pq2-4fvc) - **Mar 27, 2026**: handlebars: JavaScript injection via AST type confusion (critical). Fixed in 4.7.9. [GitHub Advisory Database](https://github.com/advisories/GHSA-2w6w-674q-4c4q) - **Mar 26, 2026**: n8n: prototype pollution in XML and GSuiteAdmin node parameters lead to RCE (critical). Fixed in 2.14.1 and 2.13.3. [GitHub Advisory Database](https://github.com/advisories/GHSA-mxrg-77hm-89hv) - **Mar 26, 2026**: langflow: authenticated code execution in Agentic Assistant validation (critical). Fixed in 1.9.0. [GitHub Advisory Database](https://github.com/advisories/GHSA-v8hw-mh8c-jxfc) - **Mar 25, 2026**: n8n: multiple remote code execution vulnerabilities in Merge node AlaSQL SQL… (critical). Fixed in 2.14.1 and 2.13.3. [GitHub Advisory Database](https://github.com/advisories/GHSA-58qr-rcgv-642v) - **Mar 25, 2026**: litellm: two LiteLLM versions published containing credential harvesting malware (critical). [GitHub Advisory Database](https://github.com/advisories/GHSA-5mg7-485q-xm76) - **Mar 19, 2026**: langflow: arbitrary file write (RCE) via v2 API (critical). Fixed in 1.9.0. [GitHub Advisory Database](https://github.com/advisories/GHSA-g2j9-7rj2-gm6c) - **Mar 17, 2026**: jspdf: HTML injection in New Window paths (critical). Fixed in 4.2.1. [GitHub Advisory Database](https://github.com/advisories/GHSA-wfv2-pwc8-crg5) - **Mar 16, 2026**: authlib: signature verification bypass (critical). Fixed in 1.6.9. [GitHub Advisory Database](https://github.com/advisories/GHSA-wvwj-cvrp-7pv5) - **Mar 10, 2026**: simple-git: blockUnsafeOperationsPlugin bypass (critical). Fixed in 3.32.3. [GitHub Advisory Database](https://github.com/advisories/GHSA-r275-fr43-pm7q) ## Reached end of life (2) - **Mar 31, 2026**: Ruby 3.2 reached end of life. No more security fixes. Latest: Ruby 4.0. [endoflife.date](https://endoflife.date/ruby) - **Mar 12, 2026**: Laravel 11 reached end of life. No more security fixes. Latest: Laravel 13. [endoflife.date](https://endoflife.date/laravel) ## New release lines (1) - **Mar 17, 2026**: Laravel 13 released. Latest version: 13.37.0. [endoflife.date](https://endoflife.date/laravel) ## Which models can’t know this Every model below has a knowledge cutoff before March 2026, so it can’t know about any of this unless a tool gives it live data. - GPT-5.6 Sol (OpenAI): February 16, 2026 - Claude Fable 5 (Anthropic): January 2026 - Claude Opus 4.7 (Anthropic): January 2026 - Claude Opus 4.8 (Anthropic): January 2026 - Claude Sonnet 5 (Anthropic): January 2026 - Grok 4.6 (xAI): January 2026 - MAI-Code-1.1-Flash (Microsoft): December 2025 - GPT-5.5 (OpenAI): December 1, 2025 - Claude Sonnet 4.6 (Anthropic): August 2025 - GPT-5.3-Codex (OpenAI): August 31, 2025 - GPT-5.4 (OpenAI): August 31, 2025 - GPT-5.4 Mini (OpenAI): August 31, 2025 - Claude Opus 4.5 (Anthropic): May 2025 - Claude Opus 4.6 (Anthropic): May 2025 - Claude Haiku 4.5 (Anthropic): February 2025 - Claude Sonnet 4.5 (Anthropic): January 2025 - Gemini 2.5 Flash (Google): January 2025 - Gemini 2.5 Pro (Google): January 2025 - Gemini 3 Flash (Google): January 2025 - Gemini 3.1 Pro (Google): January 2025 - GPT-5 (OpenAI): September 30, 2024 - GPT-5-Codex (OpenAI): September 30, 2024 - GPT-5.1 (OpenAI): September 30, 2024 - Llama 4 Maverick (Meta): August 2024 - GPT-4.1 (OpenAI): June 1, 2024 - o3 (OpenAI): June 1, 2024 - GPT-5 Mini (OpenAI): May 31, 2024 - GPT-4o (OpenAI): October 1, 2023 --- Checked Oct 17, 2026 against [CISA KEV](https://www.cisa.gov/known-exploited-vulnerabilities-catalog), [GitHub Advisory Database](https://github.com/advisories) and [endoflife.date](https://endoflife.date). Full report: https://stackprism.ai/reports/2026-03