bmad-build-auto

Runs one unattended development iteration from intent through planning, implementation, and multi-layer review.

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

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? It removes the need for human supervision during a development cycle by turning a stated intent into a planned spec, implemented code change, and triaged multi-reviewer audit in a single unattended run. ## Core Features & Use Cases - Clarify and Route: Resolves invocation intent, loads planning artifacts or compiled epic context, checks version-control state, and routes to the correct workflow step. - Spec-Driven Implementation: Generates a structured spec from a template with intent contracts, code maps, and acceptance criteria, then hands it to an implementation subagent as the sole source of truth. - Multi-Layer Review: Runs parallel review layers (blind hunter, edge-case hunter, verification-gap, intent-alignment) over the staged diff, triages every finding with verdicts, and loops back to fix specs or patch code. - Use Case: Point it at a story in a stories.yaml file or a free-form feature request; it plans, implements, verifies against the spec's test matrix, reviews the diff, and commits the finished change with a done status. ## Quick Start Invoke the bmad-build-auto skill with a description of the feature or fix you want implemented and let it run the full plan-implement-review loop unattended.

Frequently Asked Questions about bmad-build-auto

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

FAQPage Schema
How do I run an unattended AI development loop on a feature request?

Invoke the skill with your intent as the prompt. It clarifies the intent, writes a spec from its template, dispatches an implementation subagent, verifies the result, and runs parallel review layers before committing the change.

How does the skill handle stories from an epic breakdown?

It supports folder+id dispatch: it reads stories.yaml, resolves the story entry, loads the epic SPEC.md and sibling story specs for cross-story context, and writes results to an id-prefixed story spec file.

What review layers does the automated code review use?

Four layers run in parallel over the staged unified diff: a blind hunter, an edge-case hunter with deletion and claims checks, a verification-gap reviewer, and an intent-alignment auditor. Every finding is verified and triaged with a verdict before action.

What happens when a review finds problems in the implementation?

Findings route to intent_gap, bad_spec, patch, or defer. Patches are auto-fixed by re-engaging the implementation subagent; bad_spec findings trigger a spec amendment and full re-implementation, capped at five loop iterations before halting as blocked.

When does the workflow halt instead of completing?

It halts with a blocked status on unclear intent, dirty version-control trees, missing specs, failed verification, unresolved intent gaps, or review loops exceeding five iterations, always writing the status and blocking condition to the spec file.