.env.example

# =============================================================================
# Telegram Code Analyzer Configuration
# =============================================================================

# Telegram Bot (Required)
TELEGRAM_BOT_TOKEN=your_telegram_bot_token
AUTHORIZED_USERS=123456789,987654321

# Admin users who can run /index command (comma-separated Telegram IDs)
# Admins automatically have access to all bot functions (no need to add to AUTHORIZED_USERS)
ADMIN_USERS=123456789

# Project to analyze (Required)
PROJECT_PATH=/path/to/your/project

# Claude CLI Settings
CLAUDE_TIMEOUT=300000

# Claude CLI Path (optional, auto-detected if not specified)
# The bot will automatically search for Claude CLI in standard locations:
# 1. npx @anthropic-ai/claude-code (recommended)
# 2. ~/.claude/local/claude (local installation)
# 3. Global 'claude' command
# Only set this if you have Claude CLI installed in a custom location
CLAUDE_CLI_PATH=
# Examples:
# CLAUDE_CLI_PATH=/usr/local/bin/claude
# CLAUDE_CLI_PATH=/custom/path/to/claude

# =============================================================================
# LLM Provider API Keys (At least one required for RAG)
# =============================================================================

# OpenAI (embeddings: text-embedding-3-large/small)
OPENAI_API_KEY=sk-...

# Google Gemini (embeddings: gemini-embedding-001)
GEMINI_API_KEY=AIza...

# Jina AI (embeddings: jina-embeddings-v3, optimized for code)
JINA_API_KEY=jina_...

# Anthropic Claude API (no embeddings support)
ANTHROPIC_API_KEY=sk-ant-...

# Perplexity (no embeddings support)
PERPLEXITY_API_KEY=pplx-...

# Default LLM provider for completions (openai, gemini, anthropic, perplexity)
DEFAULT_LLM_PROVIDER=openai

# Default embedding provider for RAG (jina, openai, gemini)
# Jina is recommended for code analysis
DEFAULT_EMBEDDING_PROVIDER=jina

# =============================================================================
# RAG Configuration
# =============================================================================

# Path to store RAG index
RAG_STORE_PATH=./rag-index

# Chunk settings (optional, defaults shown)
# RAG_CHUNK_SIZE=300
# RAG_CHUNK_OVERLAP=50
# RAG_TOP_K=15
# RAG_RERANK_TOP_K=5
# RAG_VECTOR_WEIGHT=0.3
# RAG_LLM_WEIGHT=0.7

# =============================================================================
# Documentation RAG (ai-docs/)
# =============================================================================

# The RAG pipeline automatically indexes documentation from ai-docs/ folder.
# Documents are parsed by sections (## headings) and indexed alongside code.

# Naming Conventions for ai-docs/:
#
# Folder structure:
#   ai-docs/
#   ├── prd/           # Product Requirements Documents (ТЗ)
#   │   ├── PRD-001-feature-name.md
#   │   └── PRD-002-another-feature.md
#   ├── adr/           # Architecture Decision Records
#   │   ├── ADR-001-framework-choice.md
#   │   └── ADR-002-database-design.md
#   ├── api/           # API Specifications
#   │   ├── API-auth-endpoints.md
#   │   └── API-user-management.md
#   └── notes/         # Analysis & Research Notes
#       ├── ANALYSIS-performance-review.md
#       └── RESEARCH-llm-providers.md
#
# File naming: PREFIX-short-description.md
# Prefixes: PRD-, ADR-, API-, ANALYSIS-, RESEARCH-, SPEC-, NOTES-
#
# Frontmatter (optional but recommended):
#   ---
#   type: prd | adr | api | notes
#   status: draft | review | accepted | deprecated
#   date: 2024-01-15
#   ---

# =============================================================================
# Rate Limiting (optional)
# =============================================================================

# Maximum requests per user per minute (default: 10)
# RATE_LIMIT_MAX_REQUESTS=10

# Rate limiting time window in milliseconds (default: 60000 = 1 minute)
# RATE_LIMIT_WINDOW_MS=60000

# Cleanup interval for rate limiter memory management in milliseconds (default: 300000 = 5 minutes)
# RATE_LIMIT_CLEANUP_INTERVAL_MS=300000

# =============================================================================
# Validation (optional)
# =============================================================================

# Minimum message length in characters (default: 5)
# VALIDATION_MESSAGE_MIN_LENGTH=5

# Maximum message length in characters (default: 2000)
# VALIDATION_MESSAGE_MAX_LENGTH=2000

# Maximum users to track in rate limiter memory (default: 10000)
# RATE_LIMITER_MAX_TRACKED_USERS=10000

# =============================================================================
# Display (optional)
# =============================================================================

# Max code sources to show in RAG response (default: 3)
# RAG_MAX_SOURCES_DISPLAY=3

# =============================================================================
# RAG Internal (optional)
# =============================================================================

# Maximum directory depth for file scanning (default: 20)
# RAG_MAX_DIRECTORY_DEPTH=20

# Embedding batch size for indexing (default: 10)
# RAG_EMBEDDING_BATCH_SIZE=10

# Approximate characters per token ratio (default: 4)
# TOKENS_CHARS_RATIO=4

# Claude CLI availability check timeout in ms (default: 5000)
# CLAUDE_AVAILABILITY_CHECK_TIMEOUT=5000

# =============================================================================
# Logging
# =============================================================================

# Logging Level: DEBUG, INFO, WARN, ERROR (default: INFO, production forces WARN)
# LOG_LEVEL=INFO

# NODE_ENV=development