Glossary
The words that show up across these docs, defined once. Terms are grouped so related ideas sit together.
Dybs.it
Dybs.it : The local-first desktop app for leading a hybrid team of AI coding agents through epics, stories and sprints. It runs the BMad Method as its engine. → What is Dybs.it?
Cockpit : What Dybs.it is: the single surface you lead the work from. Every epic, story and phase visible, and one click from running. (We say cockpit, not "dashboard.")
Embedded terminal : The tabbed terminal inside Dybs.it where coding agents run. → Terminal & providers
LLM provider : The AI coding CLI an agent runs on. Any CLI that works in a terminal and supports the BMad Method can drive the workflow, which is tens to hundreds of them, from frontier agents like Claude Code and Codex CLI to open-source ones like Kimi 3 and OpenCode. A smaller set gets extra handling in Dybs.it, such as detection and session resume. You can run one provider for development and a different one for review.
Worktree : A git worktree Dybs.it creates per story, so each story develops in isolation on its own branch and parallel work never shares a checkout. Worktrees are the default substrate. Next release. → Documents, git & history
Connector : A per-project link that mirrors your epics and stories into a tool your team already uses: Notion, Obsidian or Linear. → Connectors
Command Palette
: The Cmd+K quick launcher for commands and navigation.
Intake : The Dybs.it view for capturing raw ideas before they become stories, and the first-run flow that installs the BMad Method into a new project for you. Next release.
Mandate : A feature that grants your team the authority to finish work uninterrupted (autonomous task completion). (We say Mandate, not "autopilot.")
People & roles
Agile lead : You. The human who leads a hybrid team through the BMad Method, steering the agents, reviewing their work, and deciding when something is ready to move on. (We say agile lead, not "user" or "developer.")
Hybrid team : A team made of humans and AI agents working together. In the simplest case that's one agile lead plus their coding agents; at larger scale, several agile leads each running their own agent team.
Agent : A conversational BMad Method persona with a specialty and a voice: the Analyst, Product Manager, UX Designer, Architect, Engineer, Test Architect, and Technical Writer. You launch an agent as a slash command. → Meet the agents
Brian Madison : Creator of the BMad Method, which he released as 100% open-source software. The BMad Method name and brand are trademarks of BMad Code, LLC.
The method
The BMad Method (or just "the method") : The AI-native development workflow these docs teach: four phases, a team of agents, and a trail of artifacts. → What is the BMad Method?
Phase (of the method) : One of the four stages a project moves through, Analysis → Planning → Solutioning → Implementation. → Overview
Artifact
: A durable document a phase produces: a brief, a PRD, an architecture doc, a
story spec. Artifacts carry context forward so nothing is re-derived from memory.
They live under _bmad-output/.
Product brief : The Analysis-phase artifact that captures the problem, target users, goals, success metrics, scope, and risks. → Analysis
PRFAQ : A "Working Backwards" alternative to the product brief: a press-release + FAQ that stress-tests a concept before you commit. → Analysis
PRD (Product Requirements Document) : The Planning-phase artifact that says what to build and why, framed around Jobs-to-be-Done. → Planning
UX design : The Planning-phase artifact (when there's a UI): user journeys and interface specifications. → Planning
Spine (architecture spine) : The lean set of invariants the Architecture phase produces: the rules every epic and story must respect, so independently built parts stay consistent. You architect the spine and defer the rest. → Solutioning
ADR (Architecture Decision Record) : A short document recording one architectural decision and its rationale, a supporting artifact of the Solutioning phase.
Epic : A large slice of work that groups related stories toward a shared outcome.
Story : A small, independently shippable unit of work. Each story has a spec file carrying the context the engineer agent needs. → Epics & stories
Story cycle : The repeating implementation loop: create → develop → review, looping back to develop if review finds issues, then on to the next story. → Implementation
Story lifecycle / phases : The five phases a story moves through on Dybs.it: Backlog → Ready → In Progress → Review → Done. Dybs.it tracks stories by phase (there are no kanban lane columns).
Sprint plan (sprint-status.yaml)
: The Implementation-phase artifact that orders the stories the agents work
through in sequence.
Readiness check
: The gate (the first move of /bmad-sprint-planning) that confirms the PRD, UX,
architecture, and stories all align before implementation begins.
Retrospective
: The epic-end review of completed work and lessons learned, run with
/bmad-retrospective.
Quality
TEA (Test Architecture / Test Architect) : The method's quality discipline, and Murat, the agent who owns it. Risk-based: test effort goes where the risk is. → Quality & testing
Quality gate
: The advisory epic-level decision (from /bmad-testarch-trace) that reports
whether an epic is safe to close, with a verdict of PASS / CONCERNS / FAIL /
WAIVED / NOT_EVALUATED. It informs; it doesn't hard-block. An agile lead can
waive it with rationale.
ATDD (Acceptance-Test-Driven Development) : Writing acceptance tests before the code, so the definition of done is executable.
NFR (Non-Functional Requirement)
: A quality attribute like performance, security, reliability, or scalability.
Audited with /bmad-testarch-nfr.