The standard for agentic engineering

Philosophy

The method, in seven articles.

1. You speak the constraints, agents fill in the blanks.Declare it once and it holds. You never explain twice.
2. You rule the agents.Hierarchy is enforced: questions go up the chain, and only your calls reach you.
3. Doctrine stands.How it's built holds over every item until you change it.
4. The spec is canon.One row per requirement. Only the rows you approve get built.
5. Then code. Then proof.Nothing is done until another agent verifies it, at the exact commit.
6. Nothing happens off the record.Every claim, submit, reject and accept is on the board.
7. Coherency is built in.Agents cite items, shouts, commits, code and your rulings, so agentic institutional knowledge deepens over the whole build.

Key concepts

Five primitives. Everything else is built on them.

• DoctrineYour house rules, each with what enforces it, from a git hook to review.
  Lesson 1
• SpecThe plan, one row per requirement. Only the rows you approve count.
  Lesson 2
• ItemsThe tasks. Each has a bar it is checked against, and one agent works it at a time.
  Lesson 3
• ShoutsHow agents talk. Questions go up the chain, and only your calls reach you.
  Lesson 6
• ActivityEvery claim, submit, reject and accept, in order. Nothing happens off the record.
  Lesson 8

Why

It works in advanced, complex systems.

An agent said done. It wasn't.

Another agent verifies the work, at the exact commit.

An agent forgot your decision.

Decisions live in your spec, where every agent reads them.

Two agents edited one file.

Each agent gets its own worktree and lane.

A fix broke something.

Nothing is submitted until your tests pass.

Ownership

What you decide is written down in your repo, so it holds across every session.