Reference

This section provides comprehensive reference documentation for ContextRouter configuration, commands, and APIs.

Quick Links

Reference	Description
Configuration	All settings, environment variables, and TOML options
CLI Commands	Command-line interface for ingestion and RAG
Roadmap	Development status and planned features

Configuration Overview

ContextRouter uses a layered configuration system:

┌─────────────────────────────────────────────┐
│           Runtime Settings                   │  ← Highest priority
│      (passed to functions)                   │
├─────────────────────────────────────────────┤
│           Environment Variables              │
│        (VERTEX_PROJECT_ID, etc.)            │
├─────────────────────────────────────────────┤
│             settings.toml                    │
│         (project configuration)              │
├─────────────────────────────────────────────┤
│              Default Values                  │  ← Lowest priority
│          (built into package)                │
└─────────────────────────────────────────────┘

Each layer overrides the one below it.

Configuration File Structure

A typical settings.toml:

# Model configuration
[models]
default_llm = "vertex/gemini-2.0-flash"
default_embeddings = "vertex/text-embedding-004"
temperature = 0.7
max_output_tokens = 4096

# Provider credentials
[vertex]
project_id = "my-gcp-project"
location = "us-central1"

[postgres]
host = "localhost"
database = "contextrouter"

# RAG behavior
[rag]
provider = "postgres"
reranking_enabled = true
hybrid_fusion = "rrf"

# Ingestion pipeline
[ingestion.rag]
enabled = true
output_dir = "./ingestion_output"

Environment Variables

Key environment variables:

Variable	Purpose
VERTEX_PROJECT_ID	Google Cloud project ID
VERTEX_LOCATION	GCP region (e.g., us-central1)
OPENAI_API_KEY	OpenAI API key
ANTHROPIC_API_KEY	Anthropic API key
POSTGRES_PASSWORD	Database password
RAG_PROVIDER	Default retrieval backend

API Modules

Core

from contextrouter.core import get_core_config
from contextrouter.core.registry import register_connector, register_provider
from contextrouter.core.bisquit import BisquitEnvelope

Cortex (Orchestration)

from contextrouter.cortex.runners import ChatRunner, IngestionRunner
from contextrouter.cortex.graphs import rag_retrieval, rag_ingestion
from contextrouter.cortex.state import AgentState

Models

from contextrouter.modules.models import model_registry
from contextrouter.modules.models.types import ModelRequest, TextPart

Data Sources

from contextrouter.modules.connectors import WebConnector, FileConnector
from contextrouter.modules.providers import PostgresProvider, VertexProvider

RAG

from contextrouter.modules.retrieval.rag import RagPipeline
from contextrouter.modules.retrieval.rag.settings import RagRetrievalSettings

Getting Help

• GitHub Issues: Report bugs
• Discussions: Ask questions
• Discord: Join our community (link in GitHub)