BMad Community Hub
Learn how the directory came together, why we evolve it in the open, and what to expect as we continue polishing the developer experience.
This is a reflection on what it actually feels like to move from the early excitement of a greenfield idea into the messy, unpredictable reality that every project eventually becomes. It documents the questions, missteps, and prompts that surfaced while applying the BMad method in production.
You write a spec, chart a clean path, and then the real build begins. Midway through the sprint the scope shifts, a new constraint appears, and suddenly the map is outdated. You build a TODO scratchpad full of “remember to revisit later” notes that never quite find a home in the PRDs or epics. That’s the moment where greenfield optimism fades into brownfield ambiguity.
There isn’t a universal answer for when to refactor architecture, introduce a new workflow, or pause to pay down tech debt. Every decision suddenly feels case-by-case.
To stay within the framework I started cataloging my own agent prompts:
This playbook worked for me, but there’s no guarantee it translates perfectly for someone else. That’s the tension—examples matter, yet every team’s context is slightly different.
What’s still missing is a shared space where agentic engineers can publish what actually works, cross-reference outcomes, and learn from each other without guessing in the dark. Seasoned engineers may improvise prompts instinctively; everyone else could use a leg-up with a starting point they can trust.
Greenfield and Brownfield playbooks are crucial, but they’re only the first two branches. Real products sprout dozens more: migrations, handoffs, design polish, data backfills, governance. Each branch benefits from seeing how someone else navigated it—and from having peers sanity-check the approach.
bmad.directory exists so nobody has to rediscover those workflows alone. It’s a searchable hub where we collect prompts, story templates, checklists, and retrospectives that have actually shipped. The goal is to make agentic engineering feel less like stumbling through a dark room and more like following a trail of reliable markers.
The ultimate aim is to make BMad approachable for non-developers, too—pairing these guides with community support so it grows into a universal framework anyone can use while helping each other level up.
If you’ve found a pattern that keeps your projects on track, share it. If you’re hunting for an answer, borrow what’s here, adapt it, and tell us what changed. Together we can turn one person’s rough notes into the guardrails everyone wishes they had on day one.
If this kicks off, we will introduce the Prompt Chain Builder—a block-based editor where you can compose, save, and share reusable prompt playbooks. Expect:
Once this ships, the “guardrails” concept expands from stories and checklists into living prompt flows. You’ll be able to remix someone else’s chain, annotate it with your context, and contribute it back so the next engineer has a clearer starting point.