Instant Approve auto-handles file edits, terminal commands, and recovery prompts — so you can run multiple AI conversations without clicking a single button. Local-only & secure.
A beautiful, draggable HUD injected directly into your editor. Toggle rotating fleets, track idle states, and monitor agent queues instantly.
No configuration marathons. Toggle it on and your agents run free — while you stay safe.
File creation, modification, and deletion prompts are accepted instantly. Your agents never stall waiting for you.
Build, test, and deploy commands execute without interruption. Dangerous commands are blocked by the built-in safety filter.
Stuck agents? Retry prompts and error recovery dialogs are handled automatically. No more babysitting stalled sessions.
100% Local execution. No telemetry, no cloud processing, no data exfiltration. Your IP stays on your machine.
Move your mouse or type on your keyboard? The auto-rotation engine instantly pauses and waits 10 seconds after you stop before resuming — so you can read, debug, or chat without interruption.
Scans the DOM bottom-up to prioritize new commands while cleanly ignoring historical action logs and dead buttons that break other tools.
On VS Code 1.109+, the engine leverages the official PreToolUse API. Instead of just DOM scanning, it intercepts Claude Code & Copilot actions natively at the OS layer.
Designed with a Zero-Trust architecture. If an agent command is obfuscated, malformed, or parsing fails, the engine instantly returns a hard denial to keep your system safe.
AI agents cannot rewrite their own safety rules. Hardcoded path protectors explicitly block any tool trying to mysteriously modify .vscode/settings.json or .claude/.
Under the hood lies an architecture built for high-performance scale, allowing you to parallelize your workflow without crashing your editor.
Bypasses legacy polling entirely. The engine interfaces directly with deep IDE event loops to react to agent needs instantaneously, yielding blindingly fast agent-to-environment reaction times.
Intelligently distinguishes between human commands and background Copilot noise. The engine knows precisely when you take control—pausing gracefully—and when to seamlessly resume automation.
Run five agents in parallel without bottlenecks. Our self-healing routing dynamic bypasses hung AI sessions, balancing compute seamlessly without ever dropping the automation baton.
Malicious agents cannot disable their own prison. Critical workspace configurations and system hooks are mathematically locked down, rendering self-tampering attempts futile.
High-frequency DOM interaction usually destroys IDE performance. We engineered a silent interface layer that prevents UI thrashing, keeping your scrollbars motionless and your editor buttery smooth.
Fly with complete situational awareness. A non-intrusive heads-up dashboard gives you an eagle-eye view of your entire agent fleet's state, thresholds, and performance metrics in real-time.
Every autonomous action is securely logged to an auto-rotating local JSONL database. Audit exactly what your agents did overnight with zero manual overhead.
Don't configure security twice. Export your team's banned command rules into a unified policy file that syncs via Git, protecting junior developers from rogue agents.
When a command triggers the safety shield, you aren't left guessing. An elegant contextual UI lets you instantly whitelist secure commands without diving into complex JSON settings.
We've spent hundreds of hours patching the workarounds, slipstreams, and race conditions that cause lesser tools to crash your IDE.
Rogue agents can fall into infinite retry loops, freezing your machine. A built-in mathematical rate limiter (capped at 30 actions / 3s) detects runaway behaviors and trips an Emergency Stop.
A fatal flaw in basic clickers is that agents try to accept your own extension settings. Intelligent hooks guarantee that the agent halts when you adjust Instant Approve's UI.
Basic timers break when your laptop sleeps. Our OS-level Sleep/Wake Time Dilation Guard ensures that Free Tier users aren't unfairly drained of their 30 daily minutes while disconnected.
AI agents generate commands. Sometimes those commands come from poisoned context windows, hallucinated shell scripts, or prompt injection hidden in README files you didn't write.
"YOLO mode" in your IDE doesn't check what it approves. A rogue extension, a
compromised npm package, or a single malicious .cursorrules
file
can instruct your agent to exfiltrate secrets, wipe directories, or install backdoors —
and built-in auto-accept will happily click "Run" on all of it.
Instant Approve scans every command before clicking.
rm -rf,
drop database,
format c:
base64 -d | sh,
hex payloads, eval(atob(...))
python -c exec(),
node -e child_process
↑ Agent hit a poisoned .cursorrules file mid-session.
3 exfiltration attempts blocked. You never had to look up.
Three steps. No terminal commands, no config files, no restarts.
Search "Instant Approve" in the VS Code marketplace, or click the install button above. Works in Cursor and Windsurf too.
Open any folder in VS Code. The extension detects your environment automatically.
Click "Instant Approve: OFF" in your status bar. It turns ON. That's it. Your agents now run hands-free.
Sustainable software doesn't need to bleed you monthly. Pay once, use forever.
6 months of Pro Monthly = $30. Pay $29 once and own it forever.
Instant Approve works seamlessly with VS Code, Antigravity, Cursor, and Windsurf. If you use a VS Code-based ecosystem, you're good to go.
Instant Approve includes a configurable command blocklist that prevents dangerous operations
like rm -rf /, format c:, and drop database from
being auto-approved. You can add or remove commands from this list in settings. That said,
this is a productivity tool, not a security guarantee — always review what your agents
produce.
The extension pauses auto-approval and shows an upgrade prompt. Your timer resets at midnight every day, so you get another 30 minutes tomorrow. Upgrading to Pro removes the time limit entirely.
Click the Shield icon 🛡️ in your VS Code status bar. It opens the configuration menu where you can paste your license key directly. Your key activates instantly.
Yes. Each license supports up to 3 simultaneous device activations. If you need more, deactivate one device from within VS Code (Command Palette → "Instant Approve: Remove License") or from your LemonSqueezy dashboard.
Yes. Instant Approve monitors all open windows and tabs, including minimized and unfocused ones. You can have multiple agent conversations running across multiple windows simultaneously.
Absolutely. The extension comes with a default safety list, but you can add your own banned patterns in Settings → Instant Approve: Banned Commands. Any command matching your list will be blocked.
Yes. Pro licenses are cached locally and work offline for up to 72 hours. You can code on a plane or disconnected network without interruption.
No. Zero code references are sent to our servers. The extension runs entirely locally on your machine. We are 100% "Local-Only" software. The only network traffic is for license validation.
Install for free. Upgrade when your agents outrun your patience.