codevault 1.5.0 → 1.6.1

package/.env.example

@@ -1,59 +1,94 @@
-# Embedding API Configuration
-# Using Nebius AI Studio for Qwen embeddings
+# CodeVault Environment Variables Example
+# Copy this file to .env and fill in your API keys
+
+# ============================================================================
+# EMBEDDING PROVIDER (Required)
+# ============================================================================
+
+# Option 1: Nebius AI Studio + Qwen3-Embedding-8B (Recommended)
+# - 4096 dimensions, 32K context window
+# - Best quality for code embeddings
+# - Sign up: https://nebius.com/
 CODEVAULT_EMBEDDING_API_KEY=your-nebius-api-token-here
 CODEVAULT_EMBEDDING_BASE_URL=https://api.studio.nebius.com/v1
-
-# Embedding Model Configuration
-# Qwen/Qwen3-Embedding-8B (4096 dimensions, 32K context)
-# Alternatives:
-#   - text-embedding-3-large (OpenAI, 3072 dims, 8K context)
-#   - text-embedding-3-small (OpenAI, 1536 dims, 8K context)
 CODEVAULT_EMBEDDING_MODEL=Qwen/Qwen3-Embedding-8B
+CODEVAULT_EMBEDDING_DIMENSIONS=4096
+CODEVAULT_EMBEDDING_MAX_TOKENS=32000
+CODEVAULT_EMBEDDING_RATE_LIMIT_RPM=10000
+CODEVAULT_EMBEDDING_RATE_LIMIT_TPM=600000
 
-# For Ollama (local embeddings, no API key needed)
-# CODEVAULT_OLLAMA_EMBEDDING_MODEL=nomic-embed-text
+# Option 2: Ollama (Local, Free, Private)
+# - 768 dimensions, 8K context window
+# - No API key needed, runs locally
+# - Install: https://ollama.com/
+# CODEVAULT_EMBEDDING_BASE_URL=http://localhost:11434/v1
+# CODEVAULT_EMBEDDING_MODEL=nomic-embed-text
+# CODEVAULT_EMBEDDING_DIMENSIONS=768
+# CODEVAULT_EMBEDDING_MAX_TOKENS=8192
 
-# Embedding Chunking Configuration
-# Qwen supports 32K tokens per chunk
-CODEVAULT_EMBEDDING_MAX_TOKENS=32000
+# Option 3: OpenAI
+# - 3072 dimensions (large) or 1536 (small), 8K context
+# - Reliable but more expensive
+# CODEVAULT_EMBEDDING_API_KEY=sk-your-openai-key
+# CODEVAULT_EMBEDDING_BASE_URL=https://api.openai.com/v1
+# CODEVAULT_EMBEDDING_MODEL=text-embedding-3-large
+# CODEVAULT_EMBEDDING_DIMENSIONS=3072
+# CODEVAULT_EMBEDDING_MAX_TOKENS=8192
 
-# Embedding Dimensions
-# Must match your model: Qwen = 4096, OpenAI large = 3072, small = 1536, Ollama = 768
-CODEVAULT_EMBEDDING_DIMENSIONS=4096
+# ============================================================================
+# CHAT LLM PROVIDER (Required for ask feature)
+# ============================================================================
 
-# Embedding API Rate Limiting
-# Nebius AI Studio limits (adjust based on your tier)
-CODEVAULT_EMBEDDING_RATE_LIMIT_RPM=10000
-CODEVAULT_EMBEDDING_RATE_LIMIT_TPM=600000
+# Option 1: OpenRouter + Claude Sonnet 4.5 (Recommended)
+# - 200K context window
+# - Best code understanding
+# - Sign up: https://openrouter.ai/
+CODEVAULT_CHAT_API_KEY=your-openrouter-api-key-here
+CODEVAULT_CHAT_BASE_URL=https://openrouter.ai/api/v1
+CODEVAULT_CHAT_MODEL=anthropic/claude-sonnet-4.5
+CODEVAULT_CHAT_MAX_TOKENS=32000
+CODEVAULT_CHAT_TEMPERATURE=0.1
+
+# Option 2: Ollama (Local, Free, Private)
+# - 32K context window with qwen2.5-coder
+# - Good for development and testing
+# CODEVAULT_CHAT_BASE_URL=http://localhost:11434/v1
+# CODEVAULT_CHAT_MODEL=qwen2.5-coder:7b
+# CODEVAULT_CHAT_MAX_TOKENS=32000
+# CODEVAULT_CHAT_TEMPERATURE=0.1
+
+# ============================================================================
+# RERANKING API (Optional - improves search relevance by 5-10%)
+# ============================================================================
 
-# Reranking API (Optional - improves search relevance by 5-10%)
-# Novita Qwen3-Reranker - 32K context, great for code
+# Option 1: Novita AI + Qwen3-Reranker-8B (Recommended)
+# - 32K context window
+# - Optimized for code
+# - Sign up: https://novita.ai/
 CODEVAULT_RERANK_API_URL=https://api.novita.ai/openai/v1/rerank
 CODEVAULT_RERANK_API_KEY=your-novita-api-key-here
 CODEVAULT_RERANK_MODEL=qwen/qwen3-reranker-8b
 
-# Alternatives:
-# Cohere - 4K context, $25 free credits
+# Option 2: Cohere (Alternative)
+# - 4K context window
+# - $25 free credits
 # CODEVAULT_RERANK_API_URL=https://api.cohere.ai/v1/rerank
 # CODEVAULT_RERANK_API_KEY=your-cohere-key-here
 # CODEVAULT_RERANK_MODEL=rerank-english-v3.0
 
-# Jina AI - 8K context, free tier available
-# CODEVAULT_RERANK_API_URL=https://api.jina.ai/v1/rerank
-# CODEVAULT_RERANK_API_KEY=your-jina-key-here
-# CODEVAULT_RERANK_MODEL=jina-reranker-v2-base-multilingual
+# ============================================================================
+# ENCRYPTION (Optional - for sensitive codebases)
+# ============================================================================
 
-# Chat LLM Configuration (for ask_codebase feature)
-# Using OpenRouter for Claude Sonnet 4.5 (best for code Q&A)
-CODEVAULT_CHAT_API_KEY=your-openrouter-api-key-here
-CODEVAULT_CHAT_BASE_URL=https://openrouter.ai/api/v1
-CODEVAULT_CHAT_MODEL=anthropic/claude-sonnet-4.5
-CODEVAULT_CHAT_MAX_TOKENS=32000
-CODEVAULT_CHAT_TEMPERATURE=0.1
+# Generate a secure key: openssl rand -base64 32
+# CODEVAULT_ENCRYPTION_KEY=your-32-byte-base64-key-here
+
+# ============================================================================
+# ADVANCED SETTINGS (Optional)
+# ============================================================================
 
-# For Ollama Chat (local, no API key needed)
-# CODEVAULT_OLLAMA_CHAT_MODEL=llama3.1
+# Memory management
+# CODEVAULT_CACHE_CLEAR_INTERVAL=3600000
 
-# Encryption (Optional)
-# 32-byte key encoded as base64 or hex
-# CODEVAULT_ENCRYPTION_KEY=your-encryption-key-here
+# Batch processing
+# BATCH_SIZE=50