bmad-build-auto

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

Updated Jul 2, 2026
One-click install
npx skills add https://github.com/tuanpa-nhg-eng/nhg-ipms --skill bmad-build-auto-tuanpa-nhg-eng
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: bmad-build-auto
Source: https://github.com/tuanpa-nhg-eng/nhg-ipms/tree/main/.claude/skills/bmad-build-auto
Command: npx skills add https://github.com/tuanpa-nhg-eng/nhg-ipms --skill bmad-build-auto-tuanpa-nhg-eng

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Turning a feature intent or story into reviewed, verified code normally requires constant human supervision across planning, coding, and review. This Skill runs one full unattended development loop: it clarifies intent, writes a development-ready spec, implements it via a subagent, and runs layered code reviews with automated triage and repair loopbacks. ## Core Features & Use Cases - Four-step autonomous workflow: Clarify and route intent, plan a spec from a strict template, implement via a context-free subagent, then review with parallel reviewer layers (blind hunter, edge-case hunter, verification-gap, intent-alignment). - Spec-driven development: Generates a structured spec file with intent contract, code map, tasks, acceptance criteria, and review triage logs, tracking status from draft to done. - Epic and story dispatch: Supports folder+id dispatch from a stories.yaml file, compiles focused epic context from planning artifacts, and carries cross-story continuity forward. - Use Case: Point it at a story in an epic's stories.yaml; it loads the epic context, writes the spec, implements the change, reviews the diff with four reviewer subagents, fixes or defers findings, commits the result, and halts with a final status. ## Quick Start Invoke the bmad-build-auto skill with a description of the feature or fix you want implemented, or with a spec folder and story id, and let it run the full 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 story?

Invoke the skill with a story id and spec folder containing a stories.yaml file, or with free-form intent text. It routes the intent, writes a spec, implements it through a subagent, reviews the diff, and halts with a final status such as done or blocked.

How does automated code review work in this workflow?

Step 04 builds a diff from the baseline revision and runs parallel reviewer layers: a blind hunter, an edge-case hunter, a verification-gap reviewer, and an intent-alignment auditor. Findings are triaged into intent_gap, bad_spec, patch, defer, or reject, with patches auto-fixed and bad_spec triggering spec repair loopbacks.

Can I customize the review layers or implementation handoff?

Yes. The customize.toml defaults can be overridden in _bmad/custom/bmad-build-auto.toml or a user-specific variant. You can replace review layer instructions, the implementation handoff, activation steps, and persistent facts; lists append and tables merge by key.

What happens when the workflow cannot complete a story?

It halts with status blocked and a specific blocking condition, such as unclear intent, intent gap, or verification failure. The result is written back to the story spec file or a result file under implementation artifacts, including any unresolved questions.

Does this skill require git and subagent support?

Subagents are mandatory; the workflow halts if they are unavailable. Version control is used for baseline diffs, clean-tree checks, and final commits, but the workflow degrades gracefully with best-effort change detection when version control is unavailable.