Runtime security for AI agents
Antelog checks every command your AI agent wants to run and stops the dangerous ones before they run.Free · Runs locally
How it works
Every action is checked first. #
Your AI agent
Claude Code or Cursor wants to run a command.
antelog
- 01Log it Saves the command before anything runs.
- 02Check it Safety rules look for anything dangerous.
- 03Run or stop Safe: it runs $ npm test Dangerous: stopped $ DROP DATABASE prod; Claude Code asks you first. Cursor blocks it.
↺ The result goes back to your agent.
If Antelog ever fails, it blocks the action. It never lets it through.
The problem
Nothing checks your agent before it acts. #
AI agents guess, and sometimes they guess wrong. Asking them to be careful isn’t enough.
What’s at risk
- Production databases
- Cloud accounts
- Shared git branches
Example
Stopped: Claude Code asks you to allow or decline it. Cursor blocks it.
Example: Based on the PocketOS incident of April 2026. You ask the agent to fix a credential error in staging. The agent tries to delete a storage volume, which would delete your production database and its backups. Antelog stops it and asks you. The agent then fixes the wrong setting in staging only, which Antelog allows because it deletes nothing.
Stopped: Claude Code asks you to allow or decline it. Cursor blocks it.
Example: You ask the agent to delete test users. The agent tries DELETE FROM users with no filter, which would delete every user. Antelog stops it and asks you. The agent then deletes only users with test email addresses, which Antelog allows.
Stopped: Claude Code asks you to allow or decline it. Cursor blocks it.
Example: You ask the agent why login is failing. The agent tries to print your .env file, which would put your passwords and API keys into the chat. Antelog stops it and asks you. The agent then reads the example settings file, which has no real secrets, and Antelog allows it.
Why Antelog
Why use Antelog #
<50ms
Per check
Your agent doesn’t slow down.
Custom
Add your own rules
One line stops any command you choose.
+ terraform destroy
- 19 built-in safety rules View rules
- Installs in 1 minute Install
- Runs on your computer; only unknown commands get a cloud AI check
- Every decision logged, see why anything was stopped
Install
Running in 1 minute. #
Pick your agent. Three commands, no account.
-
1### Install Also adds the Claude Code plugin. Then restart open Claude Code sessions: they aren’t protected until you do.
$ curl -fsSL https://github.com/YashSarwaiya/antelog-releases/releases/latest/download/install.sh | sh -
2### Check it works Also lists open sessions that still need a restart.
$ antelog doctor -
3### See what was blocked
$ antelog blocked -
1### Install
$ curl -fsSL https://github.com/YashSarwaiya/antelog-releases/releases/latest/download/install.sh | sh -
2### Connect Cursor Then restart Cursor so it loads the hooks.
$ antelog setup cursor -
3### Check it works
$ antelog doctor
Pricing
Safety is free. Pay for fewer interruptions and team controls. #
Free
$0
Free forever. Runs on your computer. Blocks just as reliably as Pro.
- Stops dangerous commands, on your laptop and in production
- Online dashboard
- 1,000 AI checks a month per machine
- 14-day online history
Pro
$25/ month
Or $199 a year. For one person who wants more.
- Everything in Free
- 10,000 AI checks a month
- 90-day online history
- Slack and email alerts
Card needed. Cancel before day 15 and pay nothing.
Team / Enterprise
Custom
For companies.
- Everything in Pro
- AI checks: any amount
- More people, shared rules, audit log
- Custom online history
- Single sign-on and your own servers (ask us)
FAQ
Questions, answered plainly. #
How do I stop Claude Code or Cursor from running dangerous commands? #
Install Antelog. It checks every command your agent wants to run before it runs, and stops destructive ones like rm -rf on your home folder, DROP TABLE, force-pushing to main, deleting cloud databases, and reading .env files. Install is one command and needs no account.
Can it stop an AI agent from deleting my production database? #
Yes. It stops DROP TABLE, DROP DATABASE, TRUNCATE, DELETE with no WHERE, dropdb, and deleting cloud databases and volumes through cloud command-line tools or APIs: the kind of action that wiped the PocketOS production database and its backups in April 2026.
What doesn’t it protect against? #
Commands you run yourself, other programs, and agents other than Claude Code, Cursor and Codex (unless they ask through antelog check). Through MCP servers it stops delete, reset and restore calls to database, cloud and git servers, and destructive SQL, but other MCP tools are only recorded. A harmful action that matches no rule is allowed, so add your own rules for anything specific to your setup.
Can my agent still delete test data? #
Usually yes. Dropping or emptying a local test, dev or scratch database is normal agent work, so the AI check lets it through. Destructive commands aimed at production are stopped and you decide; deleting cloud databases, volumes or backups is always stopped. When Antelog isn’t sure, it asks you instead of guessing.
Does it slow my agent down? #
Rule checks run on your computer and add milliseconds. Commands no rule knows get an AI check, which usually adds about a second. You can turn it off.
Does Antelog block commands or ask me? #
In Claude Code it stops a flagged command and asks you to allow or decline it, so you stay in control. In Cursor, and when nobody is there to answer (for example claude -p), it blocks it. If Antelog itself fails, the action is always blocked.
What happens if Antelog breaks? #
The action is blocked. Your agent waits instead of running unchecked.
Can the agent turn it off? #
No. The agent can’t change Antelog’s settings, delete its record, stop it, or uninstall it. You still can.
Is my data sent anywhere? #
Only for the free AI check: a command that changes something and that no rule recognizes, plus the text of any script it runs, with passwords and keys removed, goes to Antelog’s AI check server. The server answers yes or no and does not keep the command. Your log stays on your computer unless you connect it to your online dashboard. To turn the AI check off, write the word off into ~/.antelog/hosted-ai-check.
Which agents are supported? #
Claude Code, Cursor and Codex today, plus any agent of your own through one command (antelog check). More tools later.
Is it free? #
Yes. Blocking is free and the same on every plan, on your laptop and in production. Pro ($25 a month or $199 a year) adds 10,000 AI checks a month, 90 days of history, and Slack and email alerts. Team adds more people, shared rules and an audit log.
Can I see the code? #
The code isn’t public, but everything Antelog decides is on your computer: every action and every block is written to a tamper-evident log you can read with antelog log and antelog blocked.