bmad-architecture

Create, update, and validate architecture spine documents recording cross-unit design decisions.

Updated Mar 1, 2026
One-click install
npx skills add https://github.com/Hexalith/Hexalith.Parties --skill bmad-architecture-hexalith
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: bmad-architecture
Source: https://github.com/Hexalith/Hexalith.Parties/tree/main/.agents/skills/bmad-architecture
Command: npx skills add https://github.com/Hexalith/Hexalith.Parties --skill bmad-architecture-hexalith

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) and assets (resource) components.

What problem does it solve? Teams building a system in separate parts drift apart on the decisions that must stay consistent — boundaries, state mutation, shared data ownership. This Skill captures those invariants in a terse architecture spine document so independently built units stay compatible. ## Core Features & Use Cases - Architecture spine authoring: Distills decisions into stable AD-n blocks (Binds/Prevents/Rule) with a named paradigm, stack pins, and mermaid diagrams, working from a spec, a raw idea, or an existing codebase. - Coaching or fast drafting: Elicits decisions through open-ended questioning by default, or drafts the full spine with [ASSUMPTION] tags for review when speed matters. - Deterministic review gate: Runs lint_spine.py for mechanical checks (placeholders, duplicate AD IDs, unpinned versions) plus parallel reviewer subagents before handoff. - Use Case: Point it at a brownfield repository and ask it to derive an architecture spine; it ratifies the conventions the code already shows, logs every decision to an append-only memlog, and produces ARCHITECTURE-SPINE.md. ## Quick Start Ask the assistant to create the architecture spine for your project from the current spec or codebase.

Frequently Asked Questions about bmad-architecture

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I create an architecture document from an existing codebase?

Point the skill at the repository and choose the create intent; it investigates the real code first and ratifies the conventions already present rather than inventing new ones. Decisions are logged to a memlog and distilled into ARCHITECTURE-SPINE.md at the end.

What is an architecture spine versus a full architecture document?

A spine fixes only the invariants that keep independently built units from diverging — paradigm, boundaries, state mutation rules, shared data ownership. Structural details like stack and source tree are treated as seed that the code owns once it exists.

How does the architecture review gate work?

The gate runs lint_spine.py first for mechanical checks like duplicate AD IDs, missing Binds/Prevents/Rule fields, and unpinned stack versions. Then reviewer lenses run as parallel subagents against the spine, each writing findings to a reviews subfolder.

Can I update an existing architecture spine without renumbering decisions?

Yes. Updates resume from the run's memlog, amend Rules in place, and append new AD-n entries with the next available number. Retired IDs are never reused or renumbered, so downstream references stay stable.

Does the skill work without interactive user input?

Yes, headless mode infers intent, altitude, and inputs from the activation payload, records inferences as assumptions, and still runs the full reviewer gate. It ends with a JSON status report listing the spine path, open questions, and any conflicts.