sdlc-review

Review a finished change against ten release-safety failure modes and emit cited findings as JSON.

Updated Sep 11, 2026
One-click install
npx skills add https://github.com/zhixuan312/zz-stack --skill sdlc-review-zhixuan312
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: sdlc-review
Source: https://github.com/zhixuan312/zz-stack/tree/main/catalog/sdlc/sdlc-flow/skills/sdlc-review
Command: npx skills add https://github.com/zhixuan312/zz-stack --skill sdlc-review-zhixuan312

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Before a change ships, someone who did not write it must verify it is safe to release. This Skill performs that pre-release gate: it sweeps the change against ten failure-mode categories, cites every finding with precise locations and quoted evidence, and separates pre-existing defects from regressions the change introduced. ## Core Features & Use Cases - Ten-category failure-mode sweep: Covers verification gaps, cross-reference ripples, pre-existing vs. new regressions, missing edge cases, ordering/concurrency hazards, resource cleanup gaps, backward-compatibility breaks, safety regressions, efficiency regressions, and implicit-contract assumptions. - Evidence-grounded findings: Every finding carries a file:line locator (or section locator for non-code deliverables) plus quoted material, with severity calibrated to release-safety impact. - Plan conformance check: Reads plan.md to catch work the plan never asked for and tasks the plan declared but the change skipped. - Use Case: After an execution stage finishes a code change, dispatch this Skill to produce a review.md round record and a JSON findings block the maintainer can approve or reject without re-investigating. ## Quick Start Dispatch the sdlc-review skill to review the completed change in this initiative against plan.md before it is merged.

Frequently Asked Questions about sdlc-review

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

FAQPage Schema
How do I review a code change before release with an AI agent?▼

Dispatch this Skill after the execution stage finishes. It establishes the change-set with git diff, sweeps ten failure-mode categories, and returns cited findings as one JSON block while appending the round to review.md.

What failure modes does an automated pre-release review check?▼

It checks verification gaps, cross-reference ripples, pre-existing versus new regressions, missing edge cases, ordering and concurrency hazards, resource cleanup gaps, backward-compatibility breaks, safety regressions, efficiency regressions, and implicit-contract assumptions.

Can this review non-code deliverables like reports or workflow configurations?▼

Yes. The deliverable under review may be source code, a generated report, a workflow configuration, a data pipeline definition, or a written procedure. Findings use section names, step numbers, or field paths as locators when no line numbering exists.

Does the review skill fix the defects it finds?▼

No. It is strictly read-only toward the material under review and returns findings as text. The only file it writes is its own findings record in review.md, preserving the maintainer's ability to judge each fix.

How are pre-existing defects handled in a change review?▼

Pre-existing defects stay in the findings list with preExisting set to true, separating them from release-blocking regressions the change introduced. The change-set is established first via git diff so authorship is never guessed.

What happens if no git diff is available for the review target?▼

On a non-git target or clean tree, the reviewer states this in its notes and marks findings preExisting false only where the named material itself shows the defect, rather than guessing at authorship.