bmad-code-org pushed BMAD-METHOD onto GitHub's daily trending chart on November 15, where it held 55,387 stars and gained 19 that day S¹ — a notable figure for a repository first created in April 2025 P³. The maintainers describe the project as a "Breakthrough Method for Agile AI Driven Development" S¹, though that label is self-published and no independent review has tested the claim; the star count itself is a vendor-reported figure with no error bars or third-party audit.
My read: The star count is real but it is a vanity metric, not a quality signal. What I find genuinely interesting is the trajectory: from an April 2025 tech demo P² to 55,000-plus stars and a conda-forge package P⁵ in 19 months. The v7 rewrite is the thing to watch, but the "mid-November 2026, maybe sooner" ETA S¹ is a self-reported projection with no external accountability. I would not build a production pipeline on the alpha branch.
The project began as a technical demonstration of how purpose-built agile personas could generate planning documents for complex applications from a blank slate, according to the v1.0 release notes P². The initial release was April 6, 2025 P², and the repository was created on April 13, 2025 P³. By March 2026, the v6.1.0 release delivered what the maintainers described as their biggest architectural overhaul since v6, moving all workflows, agents and tasks onto a skill-based model P⁴. A conda-forge feedstock appeared on March 31, 2026 P⁵, giving the project a standard package-distribution channel alongside its GitHub installer.
The current stable release is v6.12, with source and documentation on a dedicated branch S¹. Version 7 is in preview alpha on the main branch, with the maintainers targeting mid-November 2026 and noting it might arrive sooner S¹.
Six languages and two optional modules round out the repository
The maintainers state that BMAD-METHOD is free and open source, with no paywall S¹. The repository spans Astro, CSS, HTML, JavaScript, Python and TypeScript P³, and carries the GitHub topics "agile," "ai," "context-engineering," "sdlc" and "spec-driven-development" P³. It currently has 6,169 forks and 98 open issues P³.
Two optional modules extend the core. Toolsmith, described as a "Smithy," creates, converts, reviews and packages skills as well as agents and modules, and includes bmad-eval for measuring them S¹. Founder, branded as "Nadia the Founder Advisor," guides startups through the business side, from idea formation and pricing to fundraising S¹. Neither module has been independently assessed for production readiness, and no benchmark data has been released.
Persona-based agents target teams running AI coding workflows
The framework targets developers using AI coding agents, a space where demand for structured workflows is growing fast. BMAD-METHOD's persona-based approach assigns different AI agents to roles such as analyst, architect or founder advisor.
A small team evaluating BMAD-METHOD today should clone the V6.12 branch rather than main, since main carries the v7 alpha S¹. The first thing to check is whether the uv-based installer runs cleanly with Node.js 20.12 and Python 3.10, because both the stable and preview routes depend on it S¹. The stable v6.12 release sets those as minimum versions S¹. For a DevOps engineer managing CI pipelines, this means adding Node.js 20.12 and Python 3.10 as baseline requirements before any BMAD-METHOD workflow can run, and pinning the V6.12 branch rather than tracking main.
The v7 release is the next checkpoint. The maintainers say mid-November 2026, maybe sooner S¹.
Sources: S1 — bmad-code-org/BMAD-METHOD: Breakthrough Method for Agile Ai Driven Dev · P2 — Version 1.0 - The Original Vision · P3 — bmad-code-org/BMAD-METHOD · P4 — Release v6.1.0 · bmad-code-org/BMAD-METHOD · GitHub · P5 — conda-forge/bmad-method-feedstock
Written from 5 sourced items, 4 of them primary.
