bmad-build

Implements features and fixes through a spec-driven workflow with multi-layer code review.

Updated Mar 14, 2026
One-click install
npx skills add https://github.com/ArchaonHW/MingGoRTS --skill bmad-build-archaonhw
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: bmad-build
Source: https://github.com/ArchaonHW/MingGoRTS/tree/main/.agents/skills/bmad-build
Command: npx skills add https://github.com/ArchaonHW/MingGoRTS --skill bmad-build-archaonhw

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Turning a feature request or bug report into verified, reviewed code usually requires juggling planning, implementation, and review by hand. This Skill automates that pipeline: it clarifies intent, writes a bounded implementation spec, dispatches implementation to a subagent, and runs layered review passes before committing. ## Core Features & Use Cases - Spec-driven implementation: Generates a token-bounded spec file (intent, code map, tasks, acceptance criteria) that serves as the sole source of truth for the implementation subagent. - Multi-layer automated review: Runs blind-hunter, edge-case-hunter, and verification-gap reviewer subagents against the staged diff, then triages every finding into patch, spec-fix, or defer categories. - Epic and sprint integration: Compiles epic context from BMAD planning artifacts and syncs story status in sprint-status.yaml. - Use Case: A developer pastes a story link or describes a bug; the Skill investigates the codebase, writes a spec, implements the change via a subagent, reviews the diff with three independent reviewers, fixes real findings, and commits with a conventional message. ## Quick Start Ask the agent to implement a feature or fix a bug using bmad-build, for example by describing the change or providing a story identifier from your epics file.

Frequently Asked Questions about bmad-build

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

FAQPage Schema
How do I implement a feature with an AI coding workflow?

Invoke bmad-build with your feature description or story link. It investigates the codebase, writes a bounded spec with tasks and acceptance criteria, dispatches implementation to a subagent, reviews the diff with multiple reviewers, and commits the verified result.

How does automated AI code review work in this workflow?

Three review layers run against the staged unified diff: a blind hunter that finds missing handling, an edge-case hunter that traces unguarded paths, and a verification-gap reviewer that checks whether tests would catch regressions. Findings are triaged by verdict and routed to patches, spec fixes, or deferred work.

Can bmad-build resume an interrupted implementation?

Yes. It reads the spec file's status frontmatter and routes accordingly: draft resumes planning, ready-for-dev or in-progress resumes implementation, and in-review resumes the review step. The baseline commit is preserved so the diff always reflects the full change.

Does bmad-build work without subagent support?

Yes, with fallbacks. If subagents are unavailable, it implements directly from the spec and writes standalone review prompts for the user to run in separate sessions, then halts until findings are pasted back.

When should I not use bmad-build for a code change?

Skip it for obvious low-risk mechanical maintenance such as small ignore-file edits, typo-only fixes, formatting-only changes, or configuration hygiene. It is designed for features, stories, bug fixes, and meaningful changes that benefit from spec and review.