Just shipped : 30% faster policy enforcement, real-time approvals dashboard, and better error messages.

The Problem We Solve

AI agents like Claude are awesome for code generation, but they can:

😱 Read your .env files

💥 Delete entire directories

🗂️ Modify sensitive database schemas

Waymark prevents this by sitting between your agent and the MCP tools.

What's New

⚡ Performance

Policy evaluation: 30% faster

Dashboard renders: 50% lighter

New indexed database queries

🎯 Better UX

Clearer error messages with reasons

New CLI commands for testing policies

Real-time action telemetry

🔐 Security Improvements

Fixed symlink bypass issue

Transactional approval workflow

Immutable policy versions

Quick Example

{ "allowedPaths" : [ "./src/**" ], "blockedPaths" : [ ".env" , "secrets/" ], "requireApproval" : [ "./migrations/**" ], "blockedCommands" : [ "rm -rf" , "DROP TABLE" ] }

Enter fullscreen mode

Exit fullscreen mode

→ Now Claude can write code safely while sensitive files stay protected.

The…