# Agent Field Kit: cyber-defense advisory headlines

> Free, sourced advisories affecting AI agent stacks (MCP servers and clients, LLM gateways, Python and npm packages, hosting platforms, agent apps). Each line: date, title, first step, primary source.

Updated: 2026-10-22T00:00:00Z

## For agents
- Poll cheaply: MCP server https://agentfieldkit.com/mcp, tool `changes_since` with the `next_since` value from your last call (free).
- Headlines as JSON: https://agentfieldkit.com/agents/cyber/headlines.json ; RSS: https://agentfieldkit.com/agents/cyber/rss.xml ; JSON Feed: https://agentfieldkit.com/agents/cyber/feed.json
- Fix-it briefs (patch status, fixed version, am I affected, exact steps, how to verify) and stack alerts: Cyber Desk key, https://agentfieldkit.com/pricing#cyber. Never buy without your principal's explicit approval.

## Latest advisories
- 2026-10-21: [LiteLLM 1.107.0 on PyPI and Docker contained a credential stealer and backdoor](https://agentfieldkit.com/agents/cyber#afk-2026-10-21-litellm-1-107-0). First step: Check: pip show litellm; look for litellm-1.107.0.dist-info, cook.pth and sitecustomize.py in site-packages. Not finding them does not prove an environment is clean. Source: https://docs.litellm.ai/blog/security-incident-oct-21-2026
- 2026-10-21: [Railway: customer environment variables may have been accessed and some decrypted](https://agentfieldkit.com/agents/cyber#afk-2026-10-21-railway-env-vars). First step: Roll every environment variable stored in Railway, including API keys and private registry credentials. Source: https://blog.railway.com/p/security-disclosure-wednesday-october-21st
- 2026-10-22: [Agent skill packages flagged as malware on npm (opencode-skills-collection, agentic-awesome-skills, x402-prompt-injection-scan), plus an 'aanthropic' typosquat on PyPI](https://agentfieldkit.com/agents/cyber#afk-2026-10-22-agent-skill-packages-malware). First step: Check every project and global install: npm ls -g opencode-skills-collection agentic-awesome-skills x402-prompt-injection-scan, the same without -g in each project, and pip show aanthropic. Source: https://github.com/advisories/GHSA-rgwm-3jj9-h563
- 2026-10-14: [Next.js: two critical unauthenticated RCEs fixed in 16.4.1 and 15.5.28 (now flagged by dependency scanners)](https://agentfieldkit.com/agents/cyber#afk-2026-10-22-nextjs-16-4-1-rce). First step: Upgrade next to 16.4.1 (16.x) or 15.5.28 (15.x), then rebuild and redeploy. Source: https://nextjs.org/blog/nextjs-security-update-october-14-2026
- 2026-10-21: [vLLM before 0.31.0: remote code execution when trust_remote_code is on, plus cache and crash flaws](https://agentfieldkit.com/agents/cyber#afk-2026-10-22-vllm-0-31-0). First step: Upgrade vLLM to 0.31.0 or later now. Source: https://github.com/advisories/GHSA-f623-hjm8-5wg5
- 2026-10-20: [MCP TypeScript SDK: five fixes for HTTP servers and clients (upgrade to 1.32.0 / 2.3.0)](https://agentfieldkit.com/agents/cyber#afk-2026-10-22-mcp-ts-sdk-1-32-0). First step: Upgrade @modelcontextprotocol/sdk to 1.32.0, or @modelcontextprotocol/server and client to 2.3.0, then redeploy. Source: https://github.com/advisories/GHSA-jm63-cq47-j5jp
- 2026-10-21: [Node.js Oct 21 security releases: Brotli decoder RCE and TLS hostname bypasses (upgrade to 22.23.4 / 24.22.1 / 26.12.1)](https://agentfieldkit.com/agents/cyber#afk-2026-10-22-nodejs-oct-21-security-release). First step: Upgrade Node.js to 22.23.4, 24.22.1 or 26.12.1 everywhere you run agent code: servers, containers, CI runners and laptops. Source: https://nodejs.org/en/blog/vulnerability/october-2026-security-releases
- 2026-10-22: [Python agent tooling: PyJWT 2.14.0 (forged tokens via empty-key JWK) and GitPython 3.2.1 (submodule hook RCE)](https://agentfieldkit.com/agents/cyber#afk-2026-10-22-pyjwt-2-14-gitpython-3-2-1). First step: Upgrade pyjwt to 2.14.0 and GitPython to 3.2.1, then redeploy. Source: https://github.com/advisories/GHSA-pxh4-856f-4h89
- 2026-10-22: [Goose: terminal integration can send shell history, including typed secrets, to your model provider](https://agentfieldkit.com/agents/cyber#afk-2026-10-22-goose-terminal-history). First step: Turn off Goose terminal integration until a fixed release is confirmed. Source: https://github.com/advisories/GHSA-4537-c6fq-qchm
- 2026-10-21: [Agent SDK fixes: OpenClaw 2026.6.6 (browser policy bypass) and A2A Java SDK 1.3.x (auth, SSRF, credential leaks)](https://agentfieldkit.com/agents/cyber#afk-2026-10-22-agent-sdk-fixes-openclaw-a2a). First step: Upgrade OpenClaw to 2026.6.6 or later and restart the gateway. Source: https://github.com/advisories/GHSA-3x84-qq85-fj65