AgentSecret – See what secrets your coding agent leaked

Scan and block AI coding agents from inadvertently sending keys and sensitive data

Problem

Developers using coding agents risk having API keys and environment variables sent to external services. The Show HN 'Which of your secrets did your coding agent send?' highlights this fear, but continuous monitoring and real-time blocking are missing.

Solution

A lightweight agent monitor that intercepts network requests from coding agents, detects secret patterns (API keys, tokens), and can block or alert on high-risk requests. Includes a local dashboard for leak history.

Users

Individual devs and teams using GitHub Copilot, Cursor, Claude Code, etc.

MVP

Intercept HTTP requests from common agents; detect secrets via regex and entropy; alert before sending and log; simple local log UI.

Revenue

Open core + paid team plan (central policy, audit logs, Slack alerts) with per-seat subscription.

Why now

AI coding agents are booming but security lags; indie devs can lead while big security vendors focus on enterprises.