AI Architecture Context Project Guide

Project information

When AI agents build software, the architecture is what they build from. If it is unclear, out of date, or buried under its own history, an agent builds the wrong thing quickly and confidently. A person reviewing the work cannot tell either, because they cannot find what the design says.

This guide shows how to keep a system's architecture as a context that people and agents can both read, trust, and change. It works at the scale of one person with an AI, and grows with a team.

What you get from it:

Architecture documents a person can read, because each says what is true now before how it came to be. Decisions that can be found, because every ruling has a number and a home, whoever made it. Changes that leave a trail, from the builder who found a problem to the decision and back to the code. An architect's body of knowledge behind it: the guide reflects the principles of IASA, the International Association of Software Architects.

The guide is being built in the open. Its patterns are usable now, and the gaps it knows about are listed in the BTABoK map and in its tasks.

AI dataset claude-code claude-ai opanai-codex ai-agents rm-odp iasa iasa-btabok