Sits between coding agents and your machine, blocking destructive commands and data leaks.
What it does
On this page
Antelog is a local guard for coding agents. It inspects each action an agent wants to take on your computer, then either lets it run or refuses it.
Who it is for
Developers who let Claude Code, Cursor, Codex or a similar tool execute shell commands on a machine that also holds production keys.
How it works
One installer connects the agents it finds. For every proposed action Antelog writes a log entry first, applies its rules plus an AI check, and finally runs or blocks the command. Refusals cover dropping databases and backups, deleting cloud volumes, force-pushing a shared branch, printing .env files or SSH keys, and sending data out after the agent has read an outside page. The log is a SQLite file in your home folder, and each entry carries a fingerprint of the one before it, so antelog verify can show whether a record was edited. Secrets are redacted before anything is stored.
Pricing
Starter is $0 and blocks the same actions as the paid tiers. Pro and Team add more AI checks and a shared view for a whole team.
Notable limits
It guards AI agents only; commands you type yourself are untouched.
Source: https://antelog.dev/
