By swt-labs
Orchestrate structured AI-assisted development in Claude Code: initialize projects with planning scaffolds, generate roadmaps and phased plans from requirements, execute tasks atomically via agents, verify with QA protocols, debug scientifically, manage todos/state/progress, and run health checks.
npx claudepluginhub swt-labs/vibe-better-with-claude-code-vbwCompress a natural language file into caveman format to save input tokens. Preserves code blocks, URLs, and structure.
View and modify VBW configuration including effort profile, verification tier, and skill-hook wiring.
Investigate a bug using the Debugger agent's scientific method protocol.
Start or continue phase discussion to build context before planning.
Run health checks on VBW installation and project setup.
Apply a quick fix or small change with commit discipline. Turbo mode -- no planning ceremony.
Display all available VBW commands with descriptions and usage examples.
Set up environment, scaffold .vbw-planning, detect project context, and bootstrap project-defining files.
List pending todos from STATE.md with action hints.
Working directory:
Save session notes for next time (state auto-persists).
Switch between work profiles or create custom ones. Profiles change effort, autonomy, and verification in one go.
Working directory:
Collect diagnostic context, classify bug or feature, and file a GitHub issue.
Run standalone research by spawning Scout agent(s) for web searches and documentation lookups.
Restore project context from .vbw-planning/ state.
Working directory:
Working directory:
View, add, or manage project conventions. Shows what VBW already knows and warns about conflicts.
Add an item to the persistent backlog in STATE.md.
Cleanly remove all VBW traces from the system before plugin uninstall.
Update VBW to the latest version with automatic cache refresh.
Run human acceptance testing on completed phase work. Presents CHECKPOINT prompts one at a time.
View changelog and recent updates since your installed version.
Requirements-to-roadmap agent for project scoping, phase decomposition, and success criteria derivation.
Investigation agent using scientific method for bug diagnosis with full codebase access and persistent debug state.
Execution agent with full tool access for implementing plan tasks with atomic commits per task.
Documentation agent for READMEs, changelogs, API docs, and guides. Read access to codebase, write access for doc files only.
Planning agent that researches, decomposes phases into plans, and self-reviews in one compaction-extended session.
Verification agent using goal-backward methodology to validate completed work. Read-only (permissionMode plan). Persists verification results via write-verification.sh through Bash.
Research agent for web searches, doc lookups, and codebase scanning. Writes RESEARCH.md files directly.
Qiushi Skill: methodology skills for AI agents guided by seeking truth from facts, with Claude Code, Cursor, OpenClaw, Codex, OpenCode, and Hermes guidance.
Executes bash commands
Hook triggers when Bash tool is used
Modifies files
Hook triggers on file write and edit operations
Uses power tools
Uses Bash, Write, or Edit tools
Comprehensive UI/UX design plugin for mobile (iOS, Android, React Native) and web applications with design systems, accessibility, and modern patterns
No model invocation
Executes directly as bash, bypassing the AI model
Runs pre-commands
Contains inline bash commands via ! syntax
Bash prerequisite issue
Uses bash pre-commands but Bash not in allowed tools
No model invocation
Executes directly as bash, bypassing the AI model
Runs pre-commands
Contains inline bash commands via ! syntax
Bash prerequisite issue
Uses bash pre-commands but Bash not in allowed tools