Tag
8 articles tagged #MCP.
P.01WebMCP is a W3C proposal that lets a web page register JavaScript functions as tools an AI agent can call directly in the browser, no server-side MCP deployment required. Here's how the API works, what Chrome's origin trial supports today, and where it breaks down.
P.02Only 8.5% of public MCP servers use OAuth, and researchers watched a honeypot MCP server get hit within 48 hours of going live. Here's the practical checklist for authentication, tool scoping, and input handling before you expose one.
P.03A critical authorization bypass in n8n-MCP let one tenant read, delete, or destroy another tenant's workflow backups in multi-tenant HTTP deployments. Here's who's affected and what to patch.
P.04Malicious AI skills and poisoned CLAUDE.md files are the new supply chain attack vector. We break down the ClawdHub incident, how MCP server exploits work, and what developers must do now.
P.05EditorPickPrompt engineering is dead. Context engineering -- managing system prompts, RAG results, tool outputs, memory, and conversation history -- is the skill that matters now. Here is what changed and why.
P.06EditorPickHow we built production multi-agent systems using Claude Agent SDK and Model Context Protocol. Covers orchestrator-worker patterns, handoff strategies, error handling, and real-world architecture decisions.
P.07Anthropic launches domain-specific plugins for Claude Cowork with MCP integrations for Slack, Figma, Salesforce, and more. AI agents now work like specialized teammates across business functions.
P.08Model Context Protocol is the new standard for connecting AI to external tools. Here's a practical guide to building, deploying, and debugging MCP servers — with real code examples from production.