Works with the AI Agents you already use
Anti-REA Shield — Stop AI Reverse Engineering of Your App
Audit and harden applications against AI-assisted reverse engineering by migrating logic and implementing behavioral defense.
Free
Anti-REA Shield — Stop AI Reverse Engineering of Your App
Example session with this skill installed
Run an REA audit on the current build artifacts in ./dist and propose a hardening plan for the pricing-engine.js logic found there.
- Read your context and instructions
- Compiled the anti-rea shield
I have completed the REA exposure audit for the artifacts in ./dist. The audit identified critical business logic exposed in the client bundle that allows for TRIVIAL reproduction of your proprietary pricing model.
REA audit: pricing-app (2024-05-22)
| Finding | Location | Verdict | Cost-to-extract |
|---|---|---|---|
| Tiered discount logic | dist/main.js:1402 | EXPOSED | TRIVIAL |
| API Key (Mapbox) | dist/vendor.js:89 | EXPOSED | TRIVIAL |
| Risk score weights | dist/pricing-engine.js:42 | EXPOSED | TRIVIAL |
Extraction plan (Module 2)
| Function | Target | Contract | Effort | Kill-switch flag |
|---|---|---|---|---|
calculateDiscount | Cloudflare Worker | POST /api/v1/pricing/discount | S | FEATURE_REMOTE_PRICING |
getRiskWeights | AWS Lambda | POST /api/v1/risk/weights | M | FEATURE_REMOTE_RISK |
Hardening (Module 3)
| Tier | Action | Perf impact (measured) |
|---|---|---|
| Tier 1 | Strip source maps and minify | 0ms |
| Tier 2 | Move validateInput to WASM (AssemblyScript) | +1.2ms (Accepted) |
Detection & response (Module 4)
| Signal observed | Stage applied | Log entry (why) |
|---|---|---|
| Uniform request cadence | Stage 2: THROTTLE | Inter-arrival variance < 5ms over 50 requests |
| Honeypot endpoint hit | Stage 5: BAN | Access to /api/v1/internal/debug-values |
Residual risk
The input validation logic remains in the client via WASM to maintain sub-millisecond UI feedback. While WASM increases extraction cost to MODERATE, a determined attacker can still decompile the .wat file to understand validation constraints. This is acceptable as the core pricing formula has been moved to the server.
Next steps
- Move the
calculateDiscountfunction to the provided Cloudflare Worker stub. - Remove
*.mapfiles from the production build pipeline. - Implement the Stage 2 throttle in your Nginx configuration.
Connects securely to your tools. The creator never sees your data.
About this skill
Protecting intellectual property from AI-driven cloning requires more than simple obfuscation. This skill addresses the reality that any code running on an attacker's hardware is eventually readable. It stops treating client-side security as a binary state and instead focuses on elevating the cost of extraction until cloning becomes uneconomical. The tool performs a four-module audit of shipped artifacts to identify business logic, secrets, and algorithms exposed in bundles, binaries, or API responses.
What it does
- Exposure auditing scans build artifacts (dist, build, .next, mobile DEX) for exposed business logic and credentials with specific file-and-line evidence.
- Extraction planning generates server-side migration strategies for high-value functions, including API contract stubs and offline fallback logic.
- Client hardening applies tiered protection including WebAssembly compilation and control-flow flattening for logic that must remain local.
- Behavioral detection builds server-side response ladders to identify automated extraction patterns without invasive client-side monitoring.
How it works
- Audit phase scans production bundles for domain keywords, exposed secrets, and source maps to rank findings by cost-to-extract.
- Migration phase maps exposed functions to edge or backend targets with defined API contracts and effort estimates.
- Hardening phase implements Tier 1-3 protections for local IP hot-paths while measuring performance impact to prevent regressions.
- Response phase configures a server-side ladder (Log, Throttle, Challenge, Cooldown) based on behavioral signals like uniform request cadence.
Frameworks & tools
Works with any stack producing client-side artifacts including React/Next.js, Electron, AssemblyScript, Rust (WASM), and mobile (Android/iOS). Uses standard CLI tools like strings, asar, and jadx for artifact inspection.
Why this beats prompting it yourself
Generic prompts often suggest brittle, easily-bypassed "anti-debugging" tricks or invasive client scanning. This skill enforces a "evidence-first" rule requiring specific extraction scenarios and prioritizes moving value to the server, which is the only provable way to prevent cloning.
Use cases
- Auditing a SaaS frontend before a public launch to ensure pricing and risk logic are not exposed in JS bundles.
- Hardening desktop or mobile apps against AI-assisted decompilation of core proprietary algorithms.
- Designing API rate-limiting and poisoning strategies to stop competitors from scraping internal model outputs.
- Identifying and removing leaked environment variables or internal source maps from production environments.
Known limitations
Does not support hardware-level enclaves or homomorphic encryption. Cannot protect code once it is decrypted in RAM for execution. Requires read access to build artifacts and source code.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Security.
Cannot guarantee 100% immunity. Detection is behavioral and server-side only. Does not support TEE/SGX or homomorphic encryption. Requires manual performance benchmarking for Tier 2+ hardening.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Free to use
No account required to browse
Trust & safety
Security scan
Security scan passed 19 days ago
- Free with an Agensi account