BMad Update: Web Bundles for Gemini & ChatGPT, plus BMM v6.8.0

Seven weeks, seven BMad Method releases (v6.3 → v6.8), four BMad Builder releases, seven Test Architect releases, and one new front door for everyone who does not live in an IDE.

🚀 Web Bundles v1.0.0 (today)

bmadcode.com/web-bundles: BMad planning skills, packaged as Google Gemini Gems and ChatGPT Custom GPTs. Six bundles in the launch shelf:

• Brainstorming Coach, Product Brief Coach, PRFAQ Coach, PRD Coach, UX Coach, Market & Industry Research

Setup is a few minutes per bundle (download, unzip, create the Gem or GPT, follow the inline steps). No IDE, no CLI, no code.

Two things worth calling out:

• Built for non-technicals AND engineers. Founders, PMs, designers, and ops staff get into the BMad lifecycle without learning an editor. Engineers running BMad in an IDE get a serious cost saver: planning on a flat-rate Gemini Advanced or ChatGPT Plus subscription, IDE budget reserved for codebase work.
• Full schema parity with the IDE skills. Start in a Gem, hand the artifacts to engineering in Claude Code, refine in a Gem next week. Nothing breaks.

Deeper write-up on the design and the SDLC story is coming as a standalone post.

📐 BMad Method v6.3 → v6.8

Seven releases. The throughline: a new generation of planning skills that defines where the rest of the platform is headed.

• bmad-prd and bmad-brief, rewritten (v6.7.0): Create / Update / Validate intents, Fast and Coaching modes, scale-adaptive output, full customize.toml control, quality-rubric validation. bmad-prd is now the single PRD skill; deprecated PRD skills route to it.
• bmad-spec, new core skill (v6.8.0): Distills any input (brain dump, PRD, transcript, brief) into a five-field SPEC.md kernel. Optimized for downstream consumption by architecture, story, and dev skills. Supersedes bmad-distillator.
• bmad-ux total overhaul (v6.8.0, breaking): Replaces bmad-create-ux-design. Two-spine output: DESIGN.md (Google Labs spec) + EXPERIENCE.md. Named-protagonist journeys, Stitch handoff, opt-in reviewer gate.
• bmad-investigate (v6.7.0): Forensic case files with evidence-graded findings for bug triage, RCA, and unfamiliar-code exploration.
• 19 new elicitation techniques + framing category (v6.8.0): Six Thinking Hats, Delphi, Steelmanning, Abstraction Laddering, and more. Library grows from 50 to 69.
• Activation guardrails hardened across 23+ skills (v6.8.0): Closes the bug class where LLMs skipped append steps and on_complete hooks silently.
• TOML customization for everything (v6.4.0): Every agent and workflow customizable via _bmad/custom/ without forking. Companion bmad-customize skill walks you through it.
• Release channels: stable / next / pinned (v6.4.0): Per-module, switchable interactively or via CLI flags.
• 18 new agent platforms (v6.5.0): 42 supported total. Sourcegraph Amp, IBM Bob, Warp, OpenHands, Replit Agent, and more.
• Non-interactive installer (v6.6.0): --set and --list-options flags make BMad CI/Docker-friendly.
• .decision-log pattern (v6.7.0): Workflow decisions tracked across sessions for clean continuations and audits.

Full BMM changelog →

🛠️ BMad Builder v1.6 → v1.8.1

• customize.toml for module authors (v1.6.0): Same framework as BMM v6.4, exposed for authoring.
• bmad-eval-runner (v1.8.0): Isolation, dependency staging, full docs for evaluating skills before shipping.
• Workflow Builder overhaul (v1.8.0): Consolidated scanners, sharper validation feedback.
• Release pipeline with App auth + cosign signing (v1.7.0): Supply-chain integrity for module distributions.

Full BMB changelog →

🧪 Test Architect v1.15 → v1.19

• Audit confidence score (v1.18.0): Findings now carry a calibrated confidence so reviewers know what to trust.
• Catalog normalization: 13-column schema, preceded-by/followed-by rename, knowledge-index scoping.

Full TEA changelog →

🎨 Creative Intelligence Suite v0.2

• customize.toml + agent roster (v0.2.0): CIS adopts the framework.
• Simplified Chinese docs (v0.2.0).

🌍 i18n

Sync and refresh across zh-CN, vi-VN, fr-FR, and cs-CZ. BMad Ecosystem sidebar localized across all four.

🔮 What's next

• Full-stack skill streamline: Architecture, story creation, ticketing, dev, code review get the bmad-prd / bmad-brief / bmad-ux template. SPEC.md and DESIGN.md / EXPERIENCE.md become first-class inputs.
• More analysis skills: Discovery interviews, customer journey mapping, opportunity sizing, competitive teardowns.
• bmad-automator (next channel): Turns planning artifacts into ready-to-execute work. Install via npx bmad-method install and pick BMad Automator.
• bmad-method-ui (community alpha): VS Code dashboard + standalone Next.js web dashboard. github.com/bmad-code-org/bmad-method-ui.
• Web Bundles shelf growth: Subscribe at bmadcode.com for the heads-up.

Build something great.

/ BMad