Architectural thinking for AI-native builders.
AI can write code. Someone still has to think.
ContextQB is a methodology, a content corpus, an MCP server, and a course platform for people building software with AI coding agents. Principles, playbooks, audits, prompts, and longer-form guides — all of it readable on the web, addressable by URI through any MCP-aware tool, and grounded in a single thesis: agents reduce the cost of writing code, they do not eliminate the need for architectural thinking.
- contextqb.com — the methodology, the guides, the install path.
- contextqb.com/start-here — five-minute orientation.
- mcp.contextqb.com — the hosted MCP server.
- members.contextqb.com — newsletter and member content.
- courses.contextqb.com — paid course platform (in build).
- contextqb — the monorepo. Methodology content, web app, MCP server, the
context.qbformat spec, and the drift detector CLI.
ContextQB contains four distinct products in one repo:
| Product | What it is |
|---|---|
| Methodology | Principles, playbooks, audits, prompts, guides + the MCP that serves them. |
| Website | contextqb.com — the public face of the methodology. |
| context.qb | A machine-readable boot manifest format for repos + the drift detector. |
| Courses | Paid course platform (Supabase + LemonSqueezy). In build. |
- Read the principles and the playbooks.
- Plug the MCP into your agent.
- Sign up at contextqb.com/kickoff to hear when the first course goes live.
An Industrial Semiotics project.