Ryvane Journal https://ryvane.ai/journal Field notes from the Ryvane AI security team: adversarial research, LLM and agent security write-ups, and hands-on guides. en Fri, 11 Sep 2026 00:00:00 GMT https://ryvane.ai/favicon.png Ryvane Journal https://ryvane.ai/journal Beyond Prompt Injection: What Our AI Honeypots Are Seeing https://ryvane.ai/journal/ai-honeypot-traffic-trends-2026 https://ryvane.ai/journal/ai-honeypot-traffic-trends-2026 Fri, 11 Sep 2026 00:00:00 GMT Aadesh Shinde and the Ryvane team What our AI honeypots recorded in mid-2026: reconnaissance, credential hunting, model fingerprinting, relay abuse and MCP tool calls, and what the trends mean. Security AI Honeypots: What They Are and How to Build One https://ryvane.ai/journal/ai-honeypots-investigation-layer https://ryvane.ai/journal/ai-honeypots-investigation-layer Fri, 11 Sep 2026 00:00:00 GMT Aadesh Shinde What honeypots do, how AI decoys differ, and how to build one from the open-source Holiday Honeypot, plus a workflow that turns raw logs into usable findings. Security MCP Tool Poisoning Attacks: Four Live Demos and How to Defend Against Them https://ryvane.ai/journal/mcp-tool-poisoning-attacks https://ryvane.ai/journal/mcp-tool-poisoning-attacks Mon, 03 Aug 2026 00:00:00 GMT Snehasis Mukhopadhyay Four working MCP tool poisoning attacks, why they succeed, and practical defenses for description injection, tool shadowing, rug pulls, and output pollution. Security Agent Memory: Build It, Break It, Secure It https://ryvane.ai/journal/agent-memory-build-break-secure https://ryvane.ai/journal/agent-memory-build-break-secure Sun, 21 Jun 2026 00:00:00 GMT Arun Nair Build four types of AI agent memory in LangGraph, test poisoning, persistent injection, leakage and extraction, and apply defenses that hold. Security Agent Skills: Build Them, Break Them, and Secure the Supply Chain https://ryvane.ai/journal/agent-skills-supply-chain-security https://ryvane.ai/journal/agent-skills-supply-chain-security Wed, 10 Jun 2026 00:00:00 GMT Arun Nair Build and test agent skills, explore prompt injection and supply-chain attacks, and learn how to review, sandbox, and scan skills before installing them. Security Joining the Claude Partner Network as a Select partner https://ryvane.ai/journal/claude-partner-network-select-partner https://ryvane.ai/journal/claude-partner-network-select-partner Wed, 03 Jun 2026 00:00:00 GMT Arun Nair Ryvane joins the Claude Partner Network as a Select partner in the Services Track. What the partnership means for our AI security research and training. Company Introducing Ryvane, and What We're Building Next https://ryvane.ai/journal/introducing-ryvane-what-were-building-next https://ryvane.ai/journal/introducing-ryvane-what-were-building-next Mon, 01 Jun 2026 00:00:00 GMT Arun Nair Why we started Ryvane, what our AI security research and audits cover, and how Ryvane Academy will help security professionals get AI-ready. Company Building AI Agents using LangChain, MCP and A2A https://ryvane.ai/journal/building-practical-ai-agents-langchain-mcp-a2a https://ryvane.ai/journal/building-practical-ai-agents-langchain-mcp-a2a Fri, 29 May 2026 00:00:00 GMT Arun Nair Build a practical AI agent system with LangChain, LangGraph, MCP tools, and A2A delegation. A hands-on guide with code and architecture walkthroughs. Agents