Quick Start

Option A: Browser Wizard (Recommended)

1. Create a new project

   npx create-runcontext my-project && cd my-project && context setup

2. The 6-step wizard runs

   The setup wizard opens in your browser and walks you through:

   1. Connect — choose your database (auto-detected from environment, or enter a connection URL)
   2. Define — name your semantic plane (e.g. sales-analytics), set owner, choose sensitivity level
   3. Scaffold — RunContext introspects your schema and generates Bronze-tier metadata
   4. Checkpoint — review what was generated (tables, columns, types, initial descriptions)
   5. Curate — enrich toward Silver (sample values, tags, glossary links, lineage)
   6. Serve — MCP server configuration is generated

3. Curate to Gold

   Gold requires agent-driven curation — your AI agent queries the actual database to add semantic roles, golden queries, guardrails, and business rules. Copy the MCP config into your IDE (Claude Code, Cursor, or Copilot) and paste the curation prompt:

   “Read context/AGENT_INSTRUCTIONS.md and curate this semantic plane to Gold.”

   The agent connects to your database via MCP, inspects real data, and fills in everything needed for Gold: semantic roles on all fields, 3+ golden queries, guardrails, business rules, and hierarchies.

4. Serve to AI agents

   context serve --stdio

   Your MCP server is live. Add it to your AI tool’s config:

   {
     "mcpServers": {
       "runcontext": {
         "command": "npx",
         "args": ["@runcontext/cli", "serve", "--stdio"]
       }
     }
   }

Option B: Tell Your AI Agent

In Claude Code, Cursor, or any agentic coding platform:

“Install @runcontext/cli and build a semantic layer for my database.”

The agent runs context setup, introspects your database, and queries real data via MCP to curate your semantic plane. It reads the generated context/AGENT_INSTRUCTIONS.md and knows exactly what to do — all the way to Gold.

Option C: Manual Pipeline

# Install
npm install -g @runcontext/cli

# Scaffold from a database
context introspect --db postgres://user:pass@host/db --select

# Auto-enrich to Silver
context enrich --target silver --apply

# Check tier progress
context tier

# Serve via MCP
context serve --stdio

What Just Happened?

The setup wizard performed these steps:

1. Connect — connected to your database (PostgreSQL, DuckDB, MySQL, SQL Server, SQLite, Snowflake, BigQuery, ClickHouse, or Databricks)
2. Define — captured your semantic plane’s name, owner, and sensitivity level
3. Scaffold — introspected every table, generated OSI YAML with columns, types, primary keys, and governance files (Bronze tier)
4. Checkpoint — presented the scaffolded metadata for your review
5. Curate — enriched toward Silver with sample values, tags, glossary links, and lineage
6. Serve — generated MCP server configuration and AGENT_INSTRUCTIONS.md for agent-driven curation to Gold

Your metadata now lives in YAML files under context/. Edit them in the visual studio, let AI agents curate them, or modify the YAML directly.

Next Steps

• Check your tier: context tier to see Bronze/Silver/Gold scores
• Open the studio: context dev --studio to curate in the browser
• Export a blueprint: context blueprint for portable OSI YAML
• Generate docs: context site for a browsable documentation site
• Read the CLI reference for all 16 commands