spike(aidd-ui): decide what the experience review adds over the code review

未关闭
#1,122 1 条评论 0 个 reaction 已指派 0 人 在 GitHub 查看

维护者通常 1 天内回复

还没有人认领这个 Issue。

评估

难度
5/5
预计耗时
一周以上
新手友好度
35/100
Issue 类型
功能
描述清晰度
描述清楚
活跃度
活跃
技术栈
typescript

调研方向

Start with plugins/aidd-ui/skills/03-review, its references/findings.md and references/fingerprint.md, then compare its findings with aidd-dev:05-review on at least two real projects, one with a UI system and one without. Read 01-system/references/lifecycle.md and delta-path.md for the pending-delta decision. Done means a per-project findings table, an evidence-backed placement recommendation, and a specified delta outcome—or a reason none is needed—and removal steps if the review is not justified.

由索引模型根据 Issue 内容生成。

描述

Problem

aidd-ui ships 03-review, an experience review that produces evidence-pinned, nonnormative findings (PR #708). The code side already has aidd-dev review, which reads the same artefacts. It is not established that the UI review finds anything the code review misses, nor where it should sit.

Three overlaps are unresolved:

  • The review could target the UI system and the feature's UI contract (is the contract coherent and complete), or the implementation (does the code respect it), or both. The code review has to read the same artefacts either way.
  • A review that always reports something is cheap to produce and useless to read. If it repeats the same findings on every run, the fix belongs in the producing skills; it earns its place only if it finds project-specific things that change from run to run.
  • The design lifecycle implements first, reviews, then validates the delta. The case of a blocking review is undefined: what becomes of the pending delta.

A reviewer in a community lab on the plugin framed it as two review levels, and asked whether the macro audit of an app, independent of any feature, is a better shape than a per-feature review.

Scope

A spike. Run 03-review on at least two real projects (one with a UI system, one without) alongside the code review and answer:

  • Which findings only the UI review produced, which only the code review, and which both.
  • Whether the same findings come back on every run (a signal to fix the producing skill instead).
  • Where it should sit: a per-feature review, a standing audit of the system, or both.
  • What happens to a pending delta after a blocking review: kept, reworked, or dropped, and who decides.
Acceptance criteria
  • A table of findings per project, split by which review produced them.
  • A recommendation on placement (feature review, audit, both) with the evidence behind it.
  • The outcome of a pending delta after a blocking review is specified in the delta lifecycle, or the spike states why none is needed.
  • If the review is not justified, the issue lists what to remove rather than leaving it.
Prior art in this repo
  • PR #708 (draft): plugins/aidd-ui/skills/03-review, references/findings.md, references/fingerprint.md; delta lifecycle in 01-system/references/lifecycle.md and delta-path.md.
  • aidd-dev:05-review: the code review the UI review overlaps with.
  • #918 (open): acceptance evidence across real browser journeys, adjacent for rendered-browser review, which PR #708 lists as follow-up.
Out of scope
  • Rendered-browser review and visual regression: tracked separately (#918, PR #708 follow-ups).
  • Changing aidd-dev review beyond reading the UI artefacts.
主要语言
TypeScript
星标
481
派生
45
平均合并
1 天 13 小时
30 天内合并 PR
91

环境准备

从这里开始

  1. 先读完整个 Issue,再读项目的贡献指南。
  2. 在 Issue 下留言说明你要接手 —— 这能避免两个人做同样的事。
  3. Fork 仓库,在一个分支上完成修改。
  4. 提交 Pull Request,并在描述里引用这个 Issue 编号。

ai-driven-dev/framework 的其他 Issue

查看 ai-driven-dev/framework 的全部 Issue

相似的 Issue

更多 TypeScript Issue

把新 issue 发到你的邮箱

精选适合新手参与的 GitHub issue 摘要。