Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/swt-labs/vibe-better-with-claude-code-vbw/llms.txt

Use this file to discover all available pages before exploring further.

Monitoring commands give you visibility into what’s been built and confidence that it meets the spec. Use them on demand — VBW also runs QA automatically via hooks during builds.

/vbw:status

Display a full project progress dashboard.
/vbw:status [--metrics]
What you get:
  • All phases with completion bars and current status
  • Velocity metrics (plans complete, phases shipped)
  • Suggested next action
Flags:
FlagDescription
--metricsAdd token consumption breakdown per agent
Example:
/vbw:status --metrics

/vbw:qa [phase]

Run deep verification on a completed phase.
/vbw:qa [phase-number] [--tier=quick|standard|deep] [--effort=level]

QA tiers

TierChecksWhen to use
Quick5–10Fast sanity check after minor changes
Standard15–25Routine phase verification
Deep30+Pre-archive or high-stakes verification
VBW selects the tier automatically based on your effort profile (thorough → deep, fast/turbo → quick). Override with --tier.

How it works

/vbw:qa uses a goal-backward methodology: it starts from the phase’s acceptance criteria and works backward to verify that each requirement is met by the implementation. This catches gaps that forward-reading checks miss. Phase auto-detection: When you omit the phase number, VBW targets the most recently completed phase. Output: Produces {NN}-VERIFICATION.md in the phase directory with pass/fail results per check. Continuous QA: Hooks run QA automatically during builds. Use /vbw:qa when you want thorough, on-demand verification — for example, before accepting a phase or before archiving. Flags:
FlagDescription
--tier=quick|standard|deepOverride the verification tier
--effort=levelOverride effort profile for this run
Example:
/vbw:qa 2 --tier=deep

/vbw:verify [phase]

Run human acceptance testing for a phase.
/vbw:verify [phase-number]

How it works

/vbw:verify presents each success criterion from the phase spec one at a time and prompts you — the human — to evaluate it. This is UAT, not automated testing. For each criterion you provide a verdict:
  • pass — criterion met
  • fail — criterion not met (with notes)
  • partial — criterion partially met (with notes)
Interruption recovery: If you stop mid-way through verification, /vbw:verify resumes from the last unevaluated criterion on the next run. Output: Produces {NN}-UAT.md with your verdicts. If any issues are found, /vbw:vibe will automatically route to UAT remediation on the next run. Phase auto-detection: When you omit the phase number, VBW targets the phase awaiting verification. Example:
/vbw:verify 2
/vbw:verify runs UAT inline in your conversation via prompts — it does not spawn a QA agent. Your judgment is the acceptance gate.

Build docs developers (and LLMs) love