DeepKeep has launched AI Lens for Developers, a security layer that watches AI coding agents and stops them from leaking data or running dangerous commands. The company announced it on October 1.

Modern coding agents can read files, run shell commands and call external tools on a developer’s machine. AI Lens monitors prompts, commands, file reads and tool calls before and after they run.

It flags leaked credentials, tokens and passwords in prompts and attached files. It detects insecure code patterns, such as functions missing authentication. Destructive commands are routed to a human for approval before they execute.

The tool supports Cursor and Claude Code now, with GitHub Copilot, OpenAI Codex, Lovable and Windsurf planned. Every session produces an audit log with device ID, prompt content and user ID.

DeepKeep says 90 per cent of developers use AI coding agents at work at least weekly. Security teams currently cannot see what these agents do on employee machines, the company says.