Skip to content

configuration.md

Latest commit

 

History

History
336 lines (265 loc) · 8.9 KB

configuration.md

File metadata and controls

336 lines (265 loc) · 8.9 KB
Raw

Configuration Guide

Agentic RAG uses a flexible configuration system that allows you to customize various aspects of the system. This guide explains how to configure the system to your needs.

Configuration File

The primary configuration method is through a JSON file, typically named config.json. This file contains settings for all components of the system.

Basic Structure

{
  "version": "1.0.0",
  "app": {
    "name": "agentic-rag",
    "environment": "development",
    "debug": true
  },
  "logging": {
    "level": "info",
    "format": "json",
    "output": "console",
    "file_path": "logs/agentic-rag.log"
  },
  "api": {
    "host": "0.0.0.0",
    "port": 8000,
    "timeout": 30,
    "cors_origins": ["*"]
  },
  "memory": {
    // Memory configuration
  },
  "planning": {
    // Planning configuration
  },
  "agents": {
    // Agents configuration
  }
}

App Configuration

The app section contains general settings for the application:

"app": {
  "name": "agentic-rag",
  "environment": "development",
  "debug": true
}
Field Type Description Default
name string Name of the application "agentic-rag"
environment string Environment ("development", "production", "test") "development"
debug boolean Enable debug mode false

Logging Configuration

The logging section controls logging behavior:

"logging": {
  "level": "info",
  "format": "json",
  "output": "console",
  "file_path": "logs/agentic-rag.log"
}
Field Type Description Default
level string Log level ("debug", "info", "warning", "error", "critical") "info"
format string Log format ("text", "json") "text"
output string Log output destination ("console", "file", "both") "console"
file_path string Path to log file (when output is "file" or "both") "logs/agentic-rag.log"

API Configuration

The api section configures the REST API:

"api": {
  "host": "0.0.0.0",
  "port": 8000,
  "timeout": 30,
  "cors_origins": ["*"]
}
Field Type Description Default
host string Host to bind to "0.0.0.0"
port number Port to listen on 8000
timeout number Request timeout in seconds 30
cors_origins array CORS allowed origins ["*"]

Memory Configuration

The memory section configures memory components:

"memory": {
  "short_term": {
    "enabled": true,
    "capacity": 1000,
    "ttl": 3600
  },
  "long_term": {
    "enabled": true,
    "connection_string": "postgresql://user:password@localhost:5432/memory_db",
    "table_name": "long_term_memory"
  }
}

Short-Term Memory

Field Type Description Default
enabled boolean Enable short-term memory true
capacity number Maximum number of entries 1000
ttl number Time-to-live in seconds (0 for no expiration) 3600

Long-Term Memory

Field Type Description Default
enabled boolean Enable long-term memory true
connection_string string Database connection string "postgresql://user:password@localhost:5432/memory_db"
table_name string Base name for memory tables "long_term_memory"

Planning Configuration

The planning section configures planning components:

"planning": {
  "react": {
    "enabled": true,
    "max_steps": 10,
    "timeout": 60
  },
  "cot": {
    "enabled": true,
    "max_depth": 5
  }
}

ReAct Planner

Field Type Description Default
enabled boolean Enable ReAct planner true
max_steps number Maximum number of steps in a plan 10
timeout number Timeout in seconds for plan creation 60

Chain of Thought Planner

Field Type Description Default
enabled boolean Enable Chain of Thought planner true
max_depth number Maximum depth of reasoning steps 5

Agents Configuration

The agents section configures agent components:

"agents": {
  "aggregator": {
    "enabled": true,
    "timeout": 30,
    "max_agents": 5
  },
  "search": {
    "enabled": true,
    "engine": "kagi",
    "api_key": "${SEARCH_API_KEY}",
    "max_results": 10
  },
  "local_data": {
    "enabled": true,
    "data_path": "./data",
    "formats": ["csv", "json", "pdf"]
  },
  "cloud": {
    "enabled": true,
    "provider": "aws",
    "region": "us-west-2",
    "credentials": {
      "access_key": "${AWS_ACCESS_KEY}",
      "secret_key": "${AWS_SECRET_KEY}"
    }
  },
  "generative": {
    "enabled": true,
    "model": "claude-3-opus-20240229",
    "api_key": "${CLAUDE_API_KEY}",
    "max_tokens": 4000,
    "temperature": 0.7
  },
  "memory_agent": {
    "enabled": true
  }
}

Aggregator Agent

Field Type Description Default
enabled boolean Enable aggregator agent true
timeout number Timeout in seconds for aggregation 30
max_agents number Maximum number of agent results to aggregate 5

Search Agent

Field Type Description Default
enabled boolean Enable search agent true
engine string Search engine to use "kagi"
api_key string API key for the search engine ""
max_results number Maximum number of results to return 10

Local Data Agent

Field Type Description Default
enabled boolean Enable local data agent true
data_path string Path to the local data directory "./data"
formats array List of supported file formats ["csv", "json", "pdf"]

Cloud Agent

Field Type Description Default
enabled boolean Enable cloud agent true
provider string Cloud provider ("aws", "azure", "gcp") "aws"
region string Cloud region "us-west-2"
credentials object Credentials for cloud access See below

Cloud Credentials

Field Type Description Default
access_key string Access key for the cloud provider ""
secret_key string Secret key for the cloud provider ""

Generative Agent

Field Type Description Default
enabled boolean Enable generative agent true
model string Language model to use "claude-3-opus-20240229"
api_key string API key for the language model service ""
max_tokens number Maximum number of tokens in generated responses 4000
temperature number Temperature parameter for generation (0.0-1.0) 0.7

Memory Agent

Field Type Description Default
enabled boolean Enable memory agent true

Environment Variables

You can use environment variables in the configuration file by using the ${VARIABLE_NAME} syntax. For example:

"api_key": "${SEARCH_API_KEY}"

This will be replaced with the value of the SEARCH_API_KEY environment variable at runtime.

Configuration Precedence

The system follows this order of precedence for configuration (from highest to lowest):

  1. Command-line arguments
  2. Environment variables
  3. Configuration file
  4. Default values

Using .env Files

You can also use a .env file to set environment variables. See .env.example for a template.

Example .env file:

SEARCH_API_KEY=your_api_key_here
AWS_ACCESS_KEY=your_aws_access_key
AWS_SECRET_KEY=your_aws_secret_key
CLAUDE_API_KEY=your_claude_api_key

Advanced Configuration

Custom Agent Configuration

To add a custom agent, add a new section to the agents object with your agent's configuration:

"my_custom_agent": {
  "enabled": true,
  "custom_param1": "value1",
  "custom_param2": "value2"
}

Then register your agent in the system by adding it to the agent initialization in app.py.

Database Options

For long-term memory, you can use different database backends by changing the connection string:

  • PostgreSQL: "postgresql://user:password@localhost:5432/memory_db"
  • SQLite: "sqlite:///memory.db"
  • MySQL: "mysql://user:password@localhost:3306/memory_db"

Troubleshooting

If you encounter issues with your configuration:

  1. Check the logs for error messages
  2. Validate your JSON syntax
  3. Ensure all required fields are present
  4. Verify that environment variables are set correctly
  5. Check file paths and permissions

For more help, see the Troubleshooting Guide.