Agentlint – A security scanner for MCP server configs

- Agentlint is an open-source tool that scans **MCP (Model Context Protocol)** server configuration files for security issues. MCP is a protocol developed by Anthropic that lets AI assistants (like Claude) connect to external tools, databases, and file systems through "servers." - As developers increasingly give AI agents access to real systems (APIs, databases, shell commands), misconfigured MCP setups create new attack surfaces — e.g., an agent could be tricked into running dangerous commands or leaking data. - Agentlint checks for problems like hardcoded secrets, overly permissive tool access, or unsafe environment variables in MCP config files (typically JSON or YAML). Think of it as a "linter" specifically for AI agent security configurations. - The project is by a solo developer (Leporis14) and lives on GitHub. It reflects a growing concern in the AI safety community: as agents gain autonomy, securing the *plumbing* between models and the outside world becomes critical.