Config File

Overview

The contextkit.config.yaml file configures your ContextKit project. It lives at the root of your project directory.

Full schema

# Directory containing your context files
context_dir: context

# Database connections for introspection, enrichment, and verification
data_sources:
  default:
    adapter: duckdb
    path: ./warehouse.duckdb
  production:
    adapter: postgres
    connection: postgres://user:pass@host:5432/analytics

Fields

context_dir

Path to the directory containing your metadata files. Defaults to context.

data_sources

Named database connections. Each source requires an adapter field plus adapter-specific configuration:

Adapter	Required Fields
duckdb	path — path to .duckdb file
postgres	connection — PostgreSQL connection URI
mysql	connection — MySQL connection URI
mssql	connection — SQL Server connection URI
sqlite	path — path to .sqlite or .sqlite3 file
snowflake	account, username, password, database, schema (optional: warehouse, role)
bigquery	project, dataset (optional: keyFilename)
clickhouse	host (optional: port)
databricks	serverHostname, httpPath, token

Examples

data_sources:
  # DuckDB
  warehouse:
    adapter: duckdb
    path: ./warehouse.duckdb

  # PostgreSQL / Neon / Supabase
  production:
    adapter: postgres
    connection: postgres://user:pass@host:5432/analytics

  # MySQL
  mysql_db:
    adapter: mysql
    connection: mysql://user:pass@host:3306/mydb

  # Snowflake
  snowflake_dwh:
    adapter: snowflake
    account: xy12345.us-east-1
    username: analyst
    password: ${SNOWFLAKE_PASSWORD}
    database: ANALYTICS
    schema: PUBLIC
    warehouse: COMPUTE_WH

  # BigQuery
  bq:
    adapter: bigquery
    project: my-gcp-project
    dataset: analytics

site

Optional configuration for the static documentation site:

Field	Default	Description
title	ContextKit	Site title shown in navigation
base_path	.	Base path for asset URLs

Using named sources

Many CLI commands accept --source <name> to specify which data source to use:

context introspect --source default
context enrich --target silver --apply --source production
context verify --source default

If you only have one source named default, commands use it automatically.

Auto-detection

The context setup wizard auto-detects databases from:

1. Existing contextkit.config.yaml
2. Environment variables (DATABASE_URL, POSTGRES_URL, etc.)
3. MCP server configs from 5 IDEs:
   • Claude Code — .claude/mcp.json
   • Cursor — ~/.cursor/mcp.json
   • VS Code — .vscode/mcp.json and VS Code settings
   • Windsurf — ~/.windsurf/mcp.json
   • Claude Desktop — ~/Library/Application Support/Claude/claude_desktop_config.json
4. Database files in the current directory (.duckdb, .sqlite, .db)