bmad-checkpoint-preview

Guides human reviewers through code changes via structured walkthrough, risk, and testing steps.

Updated Aug 3, 2026
One-click install
npx skills add https://github.com/shaahy/quicknav-desktop --skill bmad-checkpoint-preview-shaahy
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: bmad-checkpoint-preview
Source: https://github.com/shaahy/quicknav-desktop/tree/main/A%20%E6%95%99%E7%A8%8B%E9%9B%86%E5%90%88/BMAD-METHOD%E6%8A%80%E8%83%BD%E9%80%9F%E6%9F%A5/source-v6.10.0/src/bmm-skills/4-implementation/bmad-checkpoint-preview
Command: npx skills add https://github.com/shaahy/quicknav-desktop --skill bmad-checkpoint-preview-shaahy

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Reviewing a code change without context is slow and error-prone. This Skill turns a PR, commit, or branch into a guided human-in-the-loop review, walking the reviewer from intent and orientation through design walkthrough, risk surfacing, and hands-on verification before a final approve/rework decision. ## Core Features & Use Cases - Orientation & Intent Summary: Identifies the change from conversation context, sprint status, or git state, then summarizes intent and surface-area stats (files changed, modules touched, boundary crossings). - Concern-Based Walkthrough: Organizes the diff by design concerns with clickable path:line stops, generating a review trail automatically when the spec lacks one. - Risk & Testing Passes: Flags high blast-radius spots (auth, schema, billing, security) and suggests concrete manual observations of changed behavior. - Use Case: After an AI agent implements a story, invoke the checkpoint review to walk through the change concern by concern, dig into risky areas like the auth middleware, then approve the PR with confidence. ## Quick Start Ask the assistant to walk me through this change as a checkpoint review of the current branch.

Frequently Asked Questions about bmad-checkpoint-preview

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I review a code change with an AI assistant step by step?▼

Trigger the checkpoint review by saying checkpoint or walk me through this change. The workflow identifies the change from context or git state, then guides you through orientation, walkthrough, detail pass, testing, and a final decision.

How does the checkpoint review identify which change to review?▼

It checks in order: an explicit PR, commit, or branch argument; recent conversation context; sprint status files for stories in review; and current git HEAD. If none work, it asks you directly and halts after three failed exchanges.

What happens if the spec has no suggested review order?▼

The skill generates a review trail from the diff itself by reading changed files, identifying 2-5 design concerns, and selecting path:line stops. If git is unavailable, it falls back to a non-trail walkthrough mode.

Can I dig deeper into a specific area during the review?▼

Yes. Saying dig into an area like the auth changes triggers a targeted re-review that reads the relevant code in full context and checks edge cases, error handling, and race conditions, then reports findings.

Does the checkpoint review approve pull requests automatically?▼

No. The human always makes the final approve, rework, or discuss decision. For PRs it can offer to run gh pr review --approve, but only after explicit confirmation since it is a visible action on a shared resource.