# Agent Operating Rules

The daily working behavior of every AI agent contributing to Assembly Hangar.

## Always
- Read before building.
- Inspect existing files before creating new ones.
- Reuse existing patterns.
- Do not duplicate components, helpers, hooks, routes or logic.
- Do not perform large refactors unless necessary.
- Prefer small, safe, reviewable changes.
- Prioritize highest-impact work first.
- Do not get stuck on cosmetic polish while core functionality is broken.

## Required working method
1. Inspect - read relevant files, understand current structure, identify the problem.
2. Map to pyramid - locate where the task belongs before touching code.
3. Plan - smallest meaningful improvement, files to change, risk, validation steps.
4. Implement - reuse components, no duplicate logic or styling, no overengineering.
5. Validate - run available checks; confirm the page loads and the workflow still makes sense.
6. Document - update changelog and docs when behavior changes.
7. Report - what was done, why it matters, files changed, what was tested, what is open.

## After each meaningful step, report
- What was done.
- Why it matters.
- Files changed.
- Tests or checks run.
- What remains open.

A good report names the change, the rationale and what was tested. A bad report
says only "improved UI".

## Do not do this
- Do not make random screens.
- Do not create visual noise.
- Do not hide settings.
- Do not ignore workflows.
- Do not skip validation.
- Do not forget documentation.
- Do not claim done without checking.
- Do not invent architecture before inspecting the current codebase.

## Prioritization
Broken core functionality, then data save/load, then navigation and information
architecture, then settings structure, then confusing workflows, then UX
clutter, then reusable components, then documentation, then visual polish, then
mobile. Desktop-first; mobile is not the priority unless explicitly requested.