B Bumper

Your AI is about
to do something risky.

Bumper pauses it, explains what's about to happen in plain English, and waits for you to say yes or no.

npm install -g bumper-guard
Free to start — 15 bumps a month, no account needed.
Without Bumper runs blindly
$rm -rf ./public/images
Deleting…
Done.
0 questions asked images folder gone for good
Bumper catches it first
Why this exists

Vibe-coded apps are shipping with the front door left open.

When AI writes most of the code, nobody's necessarily reading every line before it runs — least of all the risky lines.

5,600
publicly deployed vibe-coded apps scanned by security researchers
400+
had real, live API keys or secrets exposed in the shipped code
2,000+
critical vulnerabilities found across the same set of apps

Source: Escape.tech's early-2026 scan of publicly deployed vibe-coded applications.

How it works

Three steps. No new habits to learn.

01

Install it once

One command wires Bumper into whichever AI coding tool you already use — no new app to open, no workflow to change.

02

It watches from inside your agent

Bumper plugs into your agent's own permission system — the same mechanism that already asks "run this command?" — not a separate app watching from the outside.

03

Risky stuff pauses for you

Safe things run instantly. Genuinely risky things stop and wait for a plain-English yes or no from you, right when it matters.

What it actually checks

The default rules, in plain terms.

Everything below runs locally out of the box. You can add your own rules — these are just the ones most vibe-coded apps get burned by.

Blocked automatically

Deleting everything, no undo

rm -rf *
Asks first

Rewriting shared history

git push --force
Asks first

Running a script from the internet, sight unseen

curl ... | sh
Asks first

Publishing your package publicly

npm publish
Asks first

Wiping data from a database

DROP TABLE ...
Asks first

A real secret key landing in your code

const key = "sk_live_..."
Where it works

One tool. Every agent you already switch between.

Most safety tools protect one agent, because they'd have to be built by the same company that makes it. Bumper is wired into all four separately, through each one's own native permission system — not a bolt-on watcher hoping to catch what leaks through.

Claude Code
Live-tested
$ rm -rf * → blocked

The first integration built — verified end to end before any other agent was added.

Cursor
Live-tested
$ rm -rf * → blocked

Verified against a real, signed-in Cursor account — not just Cursor's published hook schema.

Copilot CLI
Live-tested
$ rm -rf * → blocked

Verified against a real, GitHub-authenticated Copilot session, not a simulated one.

Codex CLI
Partial support
$ rm -rf * → Codex's own prompt asks

Codex CLI has no live hook to plug into yet, so Bumper can't add its plain-English layer here — but it still turns on Codex's built-in approval prompt. Details on the roadmap →

Pricing

Start free. Upgrade only if you outgrow it.

No account needed to install and start using Bumper. Ask-decisions — the ones Bumper actually pauses you for — are what count against the free limit; automatic allow/deny never does.

Free
€0
  • Every agent, every rule
  • 15 ask-decisions a month
  • No account needed
Pro
€7/month
  • Unlimited ask-decisions
  • Everything on the roadmap for Pro

Upgrade from inside the CLI once you've hit the limit — run bumper upgrade. No signup needed just to try Bumper.

Install & guide

Everything worth knowing before you turn this on.

Install

Same three commands regardless of which agent you use. Run the install once per project (or add --global to cover every project on the machine).

1
$npm install -g bumper-guard
$bumper start
Starts the local daemon and the approval inbox at localhost:4790. Leave this running in the background.
2
$bumper install claude-code
$bumper install cursor
$bumper install copilot
$bumper install codex
Run whichever of these match your agent — or bumper install all to wire in every one at once.
3
$bumper status
$bumper log
Check it's running, or see every decision it's made so far — automatic and yours.

How a decision actually gets made

Every risky-looking action lands in one of three buckets the instant your agent tries it.

Allow

Matches a rule marked safe — proceeds instantly, nothing shown to you.

Deny

Matches a rule marked unsafe — blocked automatically, your agent sees why.

Ask

Genuinely unclear — pauses your agent and waits up to 90 seconds for your decision. No answer in time, and it plays it safe and says no.

What to be aware of

Honest limitations, not fine print — and where we already know the fix.

If bumper start isn't running, hooks fail open — your agent's actions go through unchecked rather than getting stuck. Bumper protects you when it's on, not when it's forgotten. Fixed: bumper autostart enable runs the daemon automatically on login (Startup folder on Windows, a launchd LaunchAgent on macOS, a systemd user service on Linux) — one command, no more remembering.

Codex CLI doesn't currently expose a way for an outside tool to pause and ask before it acts. Bumper falls back to a static rule file and turns on Codex's own built-in approval prompt — real protection, just not the same plain-English experience you get everywhere else. We're watching Codex CLI for a live hook and will wire it in the moment one exists — this one isn't Bumper's call to make alone.

We tested this against a real, current install of GitHub's Copilot CLI and it works — but Copilot CLI is young and its hook schema isn't officially locked down yet, so a future update could require an adjustment. Fixed: every bumper install copilot now checks the installed Copilot CLI version against the one actually tested, and warns you the moment it drifts — instead of quietly breaking.

Privacy

Your code and commands never leave your machine.

No account needed to use Bumper. Every check runs locally against your own policy — your code and commands never leave the machine, on the free plan or Pro. Upgrading only ever sends a random device ID and a plan check, nothing about what you're actually running.

Protect the next thing you build.

One install, wired into the agent you already have open. Free for as long as we're figuring this out together.

npm install -g bumper-guard
See what's next