EdgeCase AI 🧪⚡

Adversarial Mock Data & Unit Test Fixture Generator for VS Code
Transform TypeScript types, Zod schemas, interfaces, and SQL into adversarial, edge-case test fixtures and unit tests with 1 click. Powered by Google Gemini (Free Tier), OpenAI, or 100% Offline Local Ollama.
⚡ Why EdgeCase AI?
Standard mock generators (like Faker.js) only produce happy-path placeholders ("name": "John Doe", "age": 25). They do not test what actually crashes production applications.
EdgeCase AI acts as your adversarial QA engineer. It inspects your types or schemas and generates the sneaky edge cases that trigger security bugs, memory leaks, and runtime panics:
- 💥 Unicode & Multi-byte Text Attacks: Surrogate pair emoji clusters (
👨👩👧👦), zero-width spaces, Right-To-Left (RTL) layout overrides, and strings with unescaped apostrophes (O'Connor).
- 💥 Number & Precision Hazards:
Number.MAX_SAFE_INTEGER, -0, 0.000001 IEEE-754 float precision traps, and negative balances.
- 💥 Temporal Anomalies: Leap day dates (
2024-02-29), leap second bounds (23:59:60), and extreme timezone offsets (UTC+14:00).
- 💥 Security & Injection Payloads: Polyglot SQL injection strings, client-side XSS vectors (
<script>alert(1)</script>), prototype pollution keys (__proto__, constructor), and directory traversals (../../).
- 💥 Domain Reality: Valid ICD-10 medical codes, real IATA flight codes, and realistic cross-field validations.
🚀 Features
- 1-Click Adversarial Fixtures (
.fixtures.json): Highlight any type or schema $\rightarrow$ right-click $\rightarrow$ automatically saves [filename].fixtures.json in your project with overwrite protection.
- Instant Unit Test Suites (
.test.ts): Generates complete, runnable Vitest / Jest unit test suites with describe, it, and expect assertions.
- 100% Free / Zero Token Cost (BYOK): Use Google Gemini's free API tier (millions of free tokens/month with no credit card required).
- 100% Offline & Private Local Ollama Support: Works completely offline using local models (
llama3.2, codellama, deepseek) with zero data leaving your computer.
- OpenAI Integration: Compatible with GPT-4o and GPT-4o-mini.
- Zero Overhead: No telemetry, no external server trackers, keys are encrypted inside your OS Keychain (
vscode.SecretStorage).
🛠️ Getting Started in 60 Seconds
Option A: Google Gemini (Recommended — Free & Fast Setup)
- In VS Code, press
Cmd + Shift + P (or Ctrl + Shift + P) and run:
EdgeCase AI: Configure API Key
- Select Google Gemini and paste your free API key from aistudio.google.com.
- Open any file with a TypeScript type, interface, or Zod schema.
- Highlight the code $\rightarrow$ Right-Click $\rightarrow$ "EdgeCase AI: Generate Adversarial Fixtures (JSON)"!
Option B: Local Ollama (100% Offline, Private & Free — No Key Needed)
- Ensure Ollama is running on your machine:
ollama run llama3.2
- In VS Code, press
Cmd + Shift + P and run:
EdgeCase AI: Switch AI Provider (Gemini / Ollama / OpenAI)
- Select Local Ollama.
- That's it! EdgeCase AI now generates all fixtures and tests 100% locally on your machine with zero cloud requests.
Option C: OpenAI
- Press
Cmd + Shift + P $\rightarrow$ run EdgeCase AI: Configure API Key $\rightarrow$ select OpenAI.
- Paste your OpenAI API key (
sk-...).
🔄 How to Switch AI Providers Anytime
You can toggle between Gemini, Local Ollama, and OpenAI at any time:
- Via Status Bar: Click the
$(beaker) EdgeCase AI icon in the bottom-right status bar.
- Via Command Palette: Press
Cmd + Shift + P $\rightarrow$ EdgeCase AI: Switch AI Provider.
| Feature |
Faker.js / Mock.js |
Traditional Linters |
EdgeCase AI 🧪 |
| Realistic Domain Values |
❌ (Random gibberish) |
❌ |
✅ Context-aware |
| Prototype Pollution Vectors |
❌ |
❌ |
✅ Tested automatically |
| Surrogate Pair Emoji Attacks |
❌ |
❌ |
✅ Tested automatically |
| Runnable Vitest/Jest Test File Output |
❌ |
❌ |
✅ 1-Click .test.ts creation |
| Works with Zod & TypeScript AST |
Partial |
Partial |
✅ Full native support |
| 100% Offline Local Model Mode |
✅ |
✅ |
✅ Via Ollama |
⚙️ Configuration Settings
Customize your preferences in VS Code settings.json:
{
"edgecase.provider": "gemini",
"edgecase.geminiModel": "gemini-3.6-flash",
"edgecase.openaiModel": "gpt-4o-mini",
"edgecase.fixtureCount": 5,
"edgecase.ollamaEndpoint": "http://localhost:11434",
"edgecase.ollamaModel": "llama3.2"
}
❓ Frequently Asked Questions (FAQ)
Is EdgeCase AI really free to use?
Yes! EdgeCase AI uses the Bring-Your-Own-Key (BYOK) architecture. Google AI Studio offers a completely free tier with millions of tokens per month with zero billing required. Furthermore, you can use Local Ollama completely offline for free without any API key.
Does my code get sent to third-party servers?
If you use Google Gemini or OpenAI, your highlighted code snippet is sent directly to Google or OpenAI via encrypted HTTPS. If you use Local Ollama, zero data leaves your local machine. EdgeCase AI has zero telemetry and zero intermediary proxy servers.
Will it overwrite my existing test files?
No. EdgeCase AI includes built-in overwrite protection. If a file (e.g. user.test.ts) already exists in your folder, it will prompt you whether to append, overwrite, or open in an untitled scratchpad.
🏢 Built with ❤️ by Appverse Studio
EdgeCase AI is proudly created and maintained by Appverse Studio — Engineering cutting-edge digital products, mobile apps, and developer tools.
| |