API keys
Stripe, OpenAI, AWS, GitHub — straight from your .env.
Happening in your terminal right now
Your AI assistant
just saw your API keys.
Claude Code, Codex, Gemini CLI and other AI agents read your .env, configs and logs on their own. Every secret in that context gets shipped to an LLM. LocalGuard stops it on your machine.
- 🔒 100% local
- ⚡ <50 ms per scan
- 💻 Mac · Windows · Linux
- 🚫 Zero telemetry
One ordinary coding session. Real numbers.
0secrets caught
0AI requests
0%of requests carried secrets
0 MBscanned locally
Nobody was careless. Careful people still leak.
What's slipping out
Once it's in the model,
it's out of your hands.
Tokens
Session tokens and JWTs hiding in logs and stack traces.
Passwords
DB connection strings and hard-coded credentials.
Card numbers
Test fixtures that turned out not to be test data.
Client PII
Emails and phone numbers from real customer records.
Private config
Internal hosts, endpoints and business logic.
How it works
Secrets get swapped.
Your code still works.
Without LocalGuard
STRIPE_KEY=sk_live_51Hq…x9Kz DB_URL=postgres://admin:Pr0d#2026@db OWNER=[email protected]
→ sent to the AI provider as-is
With LocalGuard
STRIPE_KEY=[SECRET_1] DB_URL=postgres://admin:[SECRET_2]@db OWNER=[EMAIL_1]
→ the model never sees the real values
- 1
Install
Lightweight desktop app built in Rust. Lives in your menu bar.
- 2
Connect in one click
Pick your AI tool, hit Connect, restart it. Done.
- 3
Code like normal
Secrets are replaced before sending — real values are restored in the replies.
Works where you already work
24 AI tools. One click.
Claude CodeCodexGemini CLIClineContinueAiderOpenCodeGooseKilo CodeQwen CodeZedOpenHandsDeepSeekCrushLettashell-gpt
Pricing
See your leaks for free.
Stop them for $5.
Detect-only
Free
- See every secret that would leak
- Live dashboard & activity log
- No account needed
Most picked
Protection
$5 first month
then $49/year + VAT · per computer
- Secrets replaced before they leave
- Real values restored in replies
- All 24 AI tool integrations
- Mac, Windows & Linux
FAQ
Questions worth answering
Does LocalGuard send my prompts anywhere?
No. Everything is scanned on your own computer. No cloud processing, no telemetry.
I never paste secrets. Why do I need this?
You don't have to paste anything. AI agents read files, logs and terminal output by themselves — keys in .env, configs and stack traces land in the context automatically.
What happens without a subscription?
The free version runs in detect-only mode: you see every leak, but nothing is blocked.
Will it slow my tools down?
No. Each scan takes under 50 ms.
Which systems are supported?
macOS (Apple Silicon), Windows x64 and Linux x64 (.AppImage and .deb).
How much is it?
$5 for the first month, then $49/year + VAT, one subscription per computer.
Stop the next leak
before it leaves.
Takes 2 minutes to set up. Your next prompt is already on its way.
Get LocalGuard — $5