skills.rest
stefanoginella
stefanoginellaCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

auto-bmad

Orchestrates the BMAD build lane end-to-end for one story or an entire epic.

30|8|Updated May 27, 2026
Search Tags:#code review#pull request#orchestration#bmad#subagent delegation#ci automation#epic pipeline
One-click install
npx skills add https://github.com/stefanoginella/auto-bmad --skill auto-bmad-stefanoginella
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: auto-bmad
Source: https://github.com/stefanoginella/auto-bmad/tree/main/auto-bmad
Command: npx skills add https://github.com/stefanoginella/auto-bmad --skill auto-bmad-stefanoginella

SYSTEM DOCUMENTATION & REQUIREMENTS

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

What problem does it solve? Running the BMAD build lane manually means repeatedly invoking plan, build, review, and test steps per story, tracking sprint status, and managing branches and PRs by hand. This Skill automates that whole pipeline with delegated subagents and human-in-the-loop checkpoints at the decisions that matter. ## Core Features & Use Cases - End-to-end story pipeline: Runs bmad-build-auto plan, build, and a follow-up review pass on a second model, plus risk-gated TEA test-architecture steps, then opens a branch and PR with a report. - Epic mode: Loops the same lane over every actionable story in an epic in one run, producing one branch, one PR, and an epic-end retrospective. - Resumable state and safety rails: Picks stories from sprint-status.yaml or a bmad-spec stories.yaml, keeps state for resume, enforces hard-stops, and never lets the orchestrator edit story code or specs directly. - Use Case: Point it at a BMAD project and run one story unattended; it plans, builds, reviews on a second model, waits on CI, and stops with a report of open questions and deferred work for your review. ## Quick Start Run /auto-bmad setup in a BMAD project to write the runtime config, then invoke /auto-bmad to plan, build, and review the next story from your sprint status.

Frequently Asked Questions about auto-bmad

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

FAQPage Schema
How do I run a whole BMAD epic in one unattended run?▼

Invoke /auto-bmad epic, optionally with --epic <N> or --spec <folder>. Epic mode loops the plan-build-review pipeline over every actionable story with no per-story halts, then opens one branch, one PR, and an epic-end retrospective.

How does auto-bmad pick which story to run next?▼

It reads the next actionable story from sprint-status.yaml, or from a bmad-spec spec folder's stories.yaml when using stories mode. You can also target a story explicitly with --story <id> or resume an interrupted run from its saved state.

Which AI coding tools does auto-bmad work with?▼

It runs on Claude Code, Codex, and opencode, re-detecting the host every run. Each pipeline phase is delegated to a subagent at a per-phase model profile, and phases can optionally be routed to an external CLI via the delegation.cli_phases config.

Does auto-bmad edit story code or specs directly?▼

No. The orchestrator only delegates and reports: every plan, build, review, and test step runs inside a delegated subagent. It never reads or edits story code and never edits the spec; it owns only git operations, state bookkeeping, and the final report.

What happens when a delegated step fails or needs a human?▼

A blocked or needs-human outcome stops the pipeline and jumps to the final report, which surfaces the blocker under a needs-human section. Hard-stop conditions like a dirty git tree, missing BMAD config, or ambiguous story targets are reported verbatim and never pushed past.

What are the prerequisites for running auto-bmad?▼

You need a BMAD project installed at version 6.11.0 or later with _bmad/config.toml present, python3 >= 3.11, uv, and the required BMAD skills including bmad-build-auto. The first run triggers a setup flow that writes the runtime config and syncs the review layers.

Related Skills

Explore More Software Engineering Skills→
co-cddoco-cddo
4

bmad-sprint-planning

Generate sprint-status.yaml from Markdown epic files and story data.

Official
Advanced
AhmedKaram50AhmedKaram50

bmad-retrospective

Guide a Scrum team through a structured Epic retrospective workflow.

Community
Advanced
SynkraAISynkraAI
3.1k

aiox-qa

Review test architecture and quality gates for software projects.

Official
Advanced
unbakedcosmosunbakedcosmos

bmad-dev-auto

Executes unattended development iterations from intent through planning, implementation, and adversarial review.

Community
Advanced