docs: prepare independent App first-use and repeat-use evaluation

[huangruiteng]

The roadmap asks for five independent first-use attempts and later repeat use, but its targets have no shared participant handout or observation sheet. This prepares one App study around GQ01: a short cash-flow request, pinned-package/source/budget freeze, independent report checks, and five unrun slots that retain failures and withdrawals.

The protocol reuses existing first-run and usage-story feedback forms. It distinguishes complete onboarding, configuration, accepted-result availability, recovery, attention and measured/unknown cost. Preparing it does not qualify an installed journey, recruit participants, or replace the routing and two-cycle team acceptance gates.

Scope and continuation

• Complete within this scope: the reusable study protocol and its EN/ZH roadmap, golden-query and adoption-loop references.
• Remaining: fill the freeze sheet after the installed product journey is qualified, then run with five consenting first-time users; repeat use stays voluntary. No package, model or participant trial ran here.
• This is a documentation-only preparation slice. Runtime, inbox, routing, telemetry and UI behavior remain outside the diff.

Validation

• Tested revision: a68d5bf
• Run state: finished
• Input classes: synthetic (repository smokes); no participant input

Check kind	Result	Evidence / limitation
static	passed	All 10 introduced Markdown links and anchors resolve, including both existing GitHub feedback forms.
static	passed	uv run --extra test python examples/public-adoption-loop-doc-smoke.py; existing workflow copy and boundary intact.
static	passed	git diff --check and changed-document public-boundary scan; five files clean.
manual	passed	Unrun ledger, fixed denominators, no maintainer-rescue pass, optional public submission/citation, unknown subscription cost, and exact accepted-version timing reviewed.
static	passed	uv run --extra test loopx canary premerge --from-git-diff; 11 selected checks and 3 direct diff checks passed; no failed checks, required skips or manual holds. Validated candidate bytes were committed without further edits.
real_entrypoint	not_applicable	No runtime or UI change; installed App and participant results remain explicitly unqualified.

Initial canary setup missed repository npm development dependencies. After installing them with npm ci --ignore-scripts, the selected semantic check was rerun; it and the full selected premerge set passed. No paid model evaluations were run.

Quality receipt: cqr_960fffc67e1c42c8e64e. One bounded review repair clarified subscription-cost coverage and excluded retrospective scoring delay. No additional runtime abstraction was needed.

Frontend / visual evidence

UI impact: none. No first viewport, navigation, documentation chrome or packaged UI changes.

Boundary

Documentation update; public docs surface. Existing roadmap S1/S4/S12/S13 and G5 acceptance are referenced without adding a milestone. Shared-authority/TS fixture impact: N/A. No private state, credentials, raw traces, internal links or machine paths are included. Every commit is DCO signed off.

[huangruiteng]

Approval conclusion (author-owned PR; GitHub blocks formal self-approval)

English verdict: APPROVE. No blocking finding in this documentation preparation slice.

Reviewed head: a68d5bf

动机

既有 roadmap 已有五人首次使用与次日复用目标，但缺少共同的参与者请求、冻结表和观察规则。本次提供可直接准备试用的完整协议，属于有价值的阶段增量；实际发行包验收和参与者试用仍未完成。

改动思路

沿用 GQ01、既有 G5 目标和 First run / Usage story 表单。先固定包、公开输入、独立答案依据、预算和五个样本，再使用普通 App 请求。详细观察留在私有研究记录，公开反馈和匿名引用分别自愿选择，不引入执行器、遥测或另一套收件箱。

具体改动

关键内容讲解

冻结表和简短双语请求让参与者不用记 Agent ID 或内部协议。观察表分别记录完整引导、就绪后配置、独立验收版本的可用时间、人工补位和已知/未知成本。五行均为未运行；失败、未参与与退出不替换。订阅成本未知不等于零，早期错误报告也不能提前停止计时。英文与中文 roadmap、golden queries 和 adoption loop 都指向同一协议。

对主干的风险

主要风险是把协议准备误读成产品已经通过验收。文档明确固定包、真实恢复与后续五人试用的缺口；维护者终端救援不计首次独立成功。实际安装、模型和参与者评测未运行，不能从文档检查推导这些结果。现有 adoption smoke、10 个新增链接/锚点、边界扫描以及风险选择的 11 项 premerge 检查和 3 项 diff 检查通过，无未解决失败、必需跳过或手工 hold。首次环境缺少 npm 开发依赖，安装后失败项与整组选中检查均通过；未查询或等待 GitHub CI。

我的整体评价

长期执行与已安装用户体验保持不变，本次只完善其评估准备。没有运行时默认、权限、兼容解码或首屏变更；状态分类、领域中立和指导/强制规则的运行时检查不适用。采用一个人工观察表，避免新增通用评分框架；有界修订已明确计时与未知成本。最终提交后的质量回执 cqr_960fffc67e1c42c8e64e 有效，远端补丁内容与本地最终 diff 一致。批准此文档范围，保持真实产品与试用验收开放；本次不合并。