EbaturanCode User Guide

What is EbaturanCode?

EbaturanCode is a privacy-first AI coding assistant for VS Code with 70+ features across security, intelligence, multi-agent orchestration, and code quality. Unlike cloud-only tools, EbaturanCode can run entirely on your machine using local AI models.

System Requirements

Installation

Option A — VSIX (Recommended):

bash
# 1. Download from ebaturan.com
# 2. Install via VS Code:
code --install-extension ebaturan-code-8.1.6.vsix

Option B — CLI:

bash
npm install -g @ebaturancode/cli
ebaturan --version

Activation & License Setup

1. Open VS Code Command Palette: Ctrl+Shift+P
2. Type: EbaturanCode: Activate License
3. Paste your license key from email
4. Status bar shows $(key) Licensed — you're ready!

Need detailed steps? See the full activation guide.

OWASP Top 10 Guard

Real-time detection of SQL injection (A03), XSS (A07), command injection, cryptographic failures (A02), and hardcoded secrets. Each finding appears as an inline diagnostic with severity level and fix suggestion.

GDPR Compliance Scanner

Detects personal data processing without consent markers. Finds email addresses, names, IP addresses, and phone numbers stored without GDPR-compliant handling.

HIPAA Compliance Scanner

First IDE extension to detect unencrypted PHI (Protected Health Information). Identifies patient data, medical records, and health information in code.

CVE Dependency Intelligence

Scans package.json and lock files against known CVE databases. Shows severity ratings inline — see exactly which dependency has which vulnerability.

OSS License Guard

Detects GPL, AGPL, SSPL, and other copyleft licenses in node_modules. Prevents accidental license violations before deployment.

Pre-commit Quality Gate

Runs 5 scanners automatically before every commit: Code Review + Security + CVE + License + Compliance. Blocks commits that fail critical checks.

Personal Code DNA

Analyzes your git history to learn your coding style: variable naming conventions, error handling patterns, async preferences, comment density. AI suggestions are then personalized to match your style.

Project Memory

Persistent project knowledge that survives context windows. Stores architectural decisions, module relationships, and team conventions. Loaded automatically for every AI interaction.

Conversation Persistence

Binary-efficient storage using MessagePack compression. Includes an inverted keyword index for fast full-text search across all past conversations.

Multi-Repo RAG

Cross-repository semantic search using BM25 TF-IDF scoring. Fully local — no vector database required. Search across multiple projects simultaneously.

How It Works

1. You submit a task via the chat panel or command palette
2. The Mayor analyzes the request and routes it to the appropriate agent(s)
3. Agents work in parallel when tasks are independent
4. Results are merged, reviewed, and presented to you with full transparency

Adaptive Model Router

EbaturanCode automatically selects the best AI model for each task type. You can override this in settings.

Supported Providers

EbaturanCode supports 500+ AI models across 50+ providers: OpenAI, Anthropic, Google Gemini, Ollama, LM Studio, Mistral, Cohere, Together AI, Fireworks, Groq, Jan, and any OpenAI-compatible API endpoint.

AI Code Review

12 quality rules enforced automatically: no empty catch blocks, no magic numbers, no 'any' types, consistent error handling, proper async/await usage, and more.

Proactive Refactor

Detects 7 anti-patterns: God File (>500 lines), Mixed Concerns, Callback Hell (>3 levels), Deep Nesting, Duplicate Logic, Primitive Obsession, Feature Envy.

Code Lens Inline

Every function annotated with: ⚠️ issue count · 🧪 test status · 📏 line count · 🔥 hotspot score. Click any annotation for details.

Architecture Drift

Save a baseline snapshot of your project structure. EbaturanCode warns you when the architecture deviates over time — prevents organic degradation.

Test Gap Detector

Finds untested public functions using static analysis only — no coverage reports needed. Prioritizes functions by complexity and call frequency.

Run-Loop Engine

Autonomous 6-phase execution cycle that replaces linear instruction execution:

Uses Kahn's algorithm for wave-based parallelization, grouping independent tasks into topological waves. Crash-safe with atomic checkpoint snapshots.

Evidence Pipeline

Mandatory proof-of-work verification before task completion. Evaluates evidence quality and diversity (screenshots, test results, build outputs, diffs). Domain-aware thresholds ensure critical tasks meet higher quality bars.

SHA-256 Code Tracking

Content-addressed storage using 256-directory sharded blob store with 2-character hash prefixes. Tracks complete file version history across sessions. Provides universal deduplication and before/after diff computation.

AI Provider Setup

Open Settings and search for "EbaturanCode" or use the command palette:

json
// settings.json
{
  "ebaturancode.apiProvider": "anthropic",
  "ebaturancode.apiKey": "sk-ant-...",
  "ebaturancode.apiModelId": "claude-sonnet-4-20250514"
}

Local AI Setup (Ollama)

bash
# 1. Install Ollama
curl -fsSL https://ollama.com/install.sh | sh

# 2. Pull a model
ollama pull qwen2.5-coder:7b

# 3. Configure EbaturanCode
{
  "ebaturancode.apiProvider": "ollama",
  "ebaturancode.ollamaBaseUrl": "http://localhost:11434",
  "ebaturancode.apiModelId": "qwen2.5-coder:7b"
}

Local AI Setup (LM Studio)

bash
# 1. Download LM Studio from lmstudio.ai
# 2. Load a model and start the local server
# 3. Configure EbaturanCode:
{
  "ebaturancode.apiProvider": "lmstudio",
  "ebaturancode.lmStudioBaseUrl": "http://localhost:1234"
}

Installation & Usage

bash
npm install -g @ebaturancode/cli

# Available commands:
ebaturan review    # AI code review on current directory
ebaturan security  # Run OWASP + CVE scan
ebaturan audit     # Full compliance audit
ebaturan memory    # View/manage project memory
ebaturan dna       # Analyze code DNA profile
ebaturan onboard   # Generate ONBOARDING.md for new devs

Examples

bash
# Review a specific file
ebaturan review src/api/payment.ts

# Security scan with output format
ebaturan security --format json --output report.json

# Generate onboarding docs
ebaturan onboard --output ONBOARDING.md

Extension not loading after install

Reload VS Code: Ctrl+Shift+P → 'Developer: Reload Window'. If still not working, check the Output panel (View → Output → EbaturanCode) for error messages.

License key shows 'invalid'

Make sure you copied the full key from email with no extra spaces. The key is a long JWT token starting with 'eyJ...'.

AI responses are slow

Check your API provider status. If using local models, ensure Ollama/LM Studio is running. Try a smaller model for faster responses.

Security scan shows false positives

Use inline comments '// ebaturan-ignore OWASP-A02' to suppress specific findings. Report persistent false positives to support.

Multi-agent mode not working

Multi-agent requires Ollama or another local provider with multiple model support. Ensure all 4 agent models are pulled.

Device limit reached

Go to the Client Portal → remove old device → activate on new device. You can self-transfer up to 5 times per month.

Still need help? Contact support@ebaturan.com — we respond within 24 hours.