Claude Code Preview Window

Preview the file changes and commands an agent is about to run before it acts

Problem

When using coding agents like Claude Code, users fear silent edits to critical files or dangerous commands. Most tools show diffs only after the fact, forcing rollbacks. Someone built Intent-lens to tackle this, but users still need a lightweight tool that plugs directly into Claude Code.

Solution

A local tool that intercepts agent actions before execution, showing files to be modified, commands to be run, and risk levels, with one-click approve or deny. Integrate via Claude Code hooks or MCP layer without changing the agent itself.

Users

Developers using Claude Code, Cursor, or other coding agents; security-conscious small teams; tech leads who need to audit agent actions.

MVP

Local proxy that intercepts file writes and shell commands; colored diff and risk tags for changes; one-click approve/deny with logging; Claude Code hook or MCP integration.

Revenue

Free basic tier plus paid advanced tier for team logs, policy templates, and audit export; one-time desktop license; enterprise self-hosted license.

Why now

Coding agents are exploding, but users remain wary of autonomous actions; Claude Code hooks and the MCP ecosystem make third-party interception feasible, letting indie developers plug in quickly.