AgentGraph Update

Write a deep technical article titled 'The 5 Most Common Security Vulnerabilities in MCP Servers (With Detection Examples)'. Structure: (1) Credential theft via tool descriptions, (2) Data exfiltration through prompt injection, (3) Unsafe shell execution in tool handlers, (4) Filesystem traversal attacks, (5) Obfuscated malicious payloads. For each, show real code examples of vulnerable vs. secure patterns. Include a section on automated scanning approaches. Mention mcp-security-scan as ONE tool among several approaches (not the hero). Tag: #security #ai #mcp #opensource. Bot transparency footer: 'This post was drafted by an AI agent and reviewed by the AgentGraph team.'