Skip to content

Configuration.md

Latest commit

 

History

History
430 lines (352 loc) · 8.22 KB

Configuration.md

File metadata and controls

430 lines (352 loc) · 8.22 KB

Configuration Reference

Complete configuration options for Rook.

Environment Variables

Server Configuration

Variable Default Description
ROOK_HOST 0.0.0.0 Server bind address
ROOK_PORT 8080 Server port
ROOK_REQUIRE_AUTH (unset) Enable API key authentication (set any value)
ROOK_API_KEY (empty) API key for authentication
RUST_LOG info Log level (trace, debug, info, warn, error)

Feature Toggles

Variable Default Description
ROOK_DISABLE_GRAPH (unset) Disable graph memory features
ROOK_DISABLE_CLASSIFICATION (unset) Disable auto-classification
ROOK_DISABLE_CONSOLIDATION (unset) Disable background consolidation
ROOK_DISABLE_INTENTIONS (unset) Disable intention checking

Provider API Keys

Variable Provider
OPENAI_API_KEY OpenAI LLM and embeddings
ANTHROPIC_API_KEY Anthropic Claude models
COHERE_API_KEY Cohere embeddings and reranking
AZURE_OPENAI_API_KEY Azure OpenAI
AZURE_OPENAI_ENDPOINT Azure OpenAI endpoint
GROQ_API_KEY Groq
TOGETHER_API_KEY Together AI
DEEPSEEK_API_KEY DeepSeek
GEMINI_API_KEY Google Gemini
HUGGINGFACE_API_KEY HuggingFace
PINECONE_API_KEY Pinecone vector store
QDRANT_API_KEY Qdrant Cloud
WEAVIATE_API_KEY Weaviate Cloud

Provider Configuration

LLM Providers

OpenAI

{
  "llm": {
    "provider": "openai",
    "model": "gpt-4-turbo",
    "api_key": "sk-...",
    "base_url": null,
    "temperature": 0.7,
    "max_tokens": 4096
  }
}

Models: gpt-4-turbo, gpt-4, gpt-4o, gpt-3.5-turbo

Anthropic

{
  "llm": {
    "provider": "anthropic",
    "model": "claude-3-opus-20240229",
    "api_key": "sk-ant-..."
  }
}

Models: claude-3-opus-20240229, claude-3-sonnet-20240229, claude-3-haiku-20240307

Ollama (Local)

{
  "llm": {
    "provider": "ollama",
    "model": "llama2",
    "base_url": "http://localhost:11434"
  }
}

Models: Any model available in your Ollama installation

Azure OpenAI

{
  "llm": {
    "provider": "azure_openai",
    "model": "gpt-4",
    "api_key": "...",
    "base_url": "https://your-resource.openai.azure.com"
  }
}

Embedding Providers

OpenAI

{
  "embedder": {
    "provider": "openai",
    "model": "text-embedding-3-small",
    "api_key": "sk-..."
  }
}

Models:

  • text-embedding-3-small (1536 dims) - Recommended
  • text-embedding-3-large (3072 dims)
  • text-embedding-ada-002 (1536 dims)

Cohere

{
  "embedder": {
    "provider": "cohere",
    "model": "embed-english-v3.0",
    "api_key": "..."
  }
}

Ollama

{
  "embedder": {
    "provider": "ollama",
    "model": "nomic-embed-text",
    "base_url": "http://localhost:11434"
  }
}

Vector Store Providers

Qdrant

{
  "vector_store": {
    "provider": "qdrant",
    "url": "http://localhost:6333",
    "api_key": null,
    "collection_name": "rook",
    "embedding_dims": 1536
  }
}

Setup:

docker run -p 6333:6333 -p 6334:6334 qdrant/qdrant

PostgreSQL + pgvector

{
  "vector_store": {
    "provider": "pgvector",
    "url": "postgresql://user:pass@localhost/rook",
    "collection_name": "memories",
    "embedding_dims": 1536
  }
}

Setup:

CREATE EXTENSION vector;

Pooled PostgreSQL

For production with connection pooling:

{
  "vector_store": {
    "provider": "pgvector_pooled",
    "url": "postgresql://user:pass@localhost/rook",
    "collection_name": "memories",
    "embedding_dims": 1536,
    "pool_size": 16
  }
}

Pinecone

{
  "vector_store": {
    "provider": "pinecone",
    "api_key": "...",
    "environment": "us-east-1-aws",
    "index_name": "rook"
  }
}

Weaviate

{
  "vector_store": {
    "provider": "weaviate",
    "url": "http://localhost:8080",
    "api_key": null,
    "collection_name": "Memory"
  }
}

Milvus

{
  "vector_store": {
    "provider": "milvus",
    "url": "http://localhost:19530",
    "collection_name": "rook"
  }
}

Graph Store Providers

Embedded (Default)

{
  "graph_store": {
    "provider": "embedded",
    "path": "./rook_graph.db"
  }
}

Uses SQLite + petgraph for zero-dependency graph storage.

Neo4j

{
  "graph_store": {
    "provider": "neo4j",
    "url": "bolt://localhost:7687",
    "username": "neo4j",
    "password": "password"
  }
}

Reranker Providers

Cohere

{
  "reranker": {
    "provider": "cohere",
    "model": "rerank-english-v2.0",
    "api_key": "..."
  }
}

LLM-based

{
  "reranker": {
    "provider": "llm"
  }
}

Uses the configured LLM for reranking.

FSRS Configuration

Memory strength parameters:

Parameter Default Description
fsrs_decay_rate 0.9 Power-law decay exponent
fsrs_stability_base 1.0 Initial stability for new memories
fsrs_difficulty_default 0.3 Default difficulty (0-1)
fsrs_forgetting_threshold 0.1 Archive when retrievability below this

Classification Configuration

Parameter Default Description
classification_enabled true Enable auto-classification
key_memory_cap 15 Max key memories per search

Categories:

  • Personal, Preference, Fact, Belief, Skill
  • Experience, Social, Work, Health, Other

Consolidation Configuration

Parameter Default Description
consolidation_interval 3600 Seconds between consolidation runs
synaptic_tag_decay 3600 Tag decay time constant (seconds)
behavioral_window_before 1800 Novelty boost window before (30 min)
behavioral_window_after 7200 Novelty boost window after (2 hours)

Intention Configuration

Parameter Default Description
intention_check_interval 10 Check every N messages
keyword_bloom_fpr 0.001 Bloom filter false positive rate
semantic_threshold 0.75 Topic match threshold

Event Configuration

Parameter Default Description
events_enabled true Emit memory lifecycle events
webhook_timeout 30 Webhook timeout (seconds)
webhook_retries 3 Retry count for failed webhooks

Retrieval Configuration

Parameter Default Description
default_retrieval_mode Standard Default search mode
spreading_decay 0.5 Activation decay per hop
spreading_threshold 0.1 Minimum activation to propagate
spreading_max_depth 3 Maximum propagation depth
dedup_threshold 0.95 Similarity threshold for deduplication

Example Configurations

Development (Minimal)

{
  "llm": {
    "provider": "openai",
    "api_key": "sk-..."
  },
  "embedder": {
    "provider": "openai",
    "api_key": "sk-..."
  },
  "vector_store": {
    "provider": "qdrant",
    "url": "http://localhost:6333"
  }
}

Production (Full)

{
  "llm": {
    "provider": "openai",
    "model": "gpt-4-turbo",
    "api_key": "sk-...",
    "temperature": 0.3
  },
  "embedder": {
    "provider": "openai",
    "model": "text-embedding-3-large",
    "api_key": "sk-..."
  },
  "vector_store": {
    "provider": "pgvector_pooled",
    "url": "postgresql://user:pass@db.example.com/rook",
    "embedding_dims": 3072,
    "pool_size": 32
  },
  "graph_store": {
    "provider": "neo4j",
    "url": "bolt://neo4j.example.com:7687",
    "username": "neo4j",
    "password": "..."
  },
  "reranker": {
    "provider": "cohere",
    "api_key": "..."
  }
}

Local/Offline

{
  "llm": {
    "provider": "ollama",
    "model": "llama2",
    "base_url": "http://localhost:11434"
  },
  "embedder": {
    "provider": "ollama",
    "model": "nomic-embed-text",
    "base_url": "http://localhost:11434"
  },
  "vector_store": {
    "provider": "qdrant",
    "url": "http://localhost:6333"
  },
  "graph_store": {
    "provider": "embedded"
  }
}