Files
2026-11-16 17:26:44 -08:00

271 lines
17 KiB
Plaintext
Raw Permalink Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
你需要操作移动设备,并按要求复现。
## 工作区
- /fixture/workspace/task-1
- 代码仓: `/fixture/ai_workspace` 下
## 任务契约
- 目标: 在连接的设备上跑到 SecuritySpec 里的指定场景,验证功能路线,并通过给定的frida hook日志、及关键功能截图,证明你成功进入了目标功能。
- 允许: 为完成观测,可以使用 Frida hook 修改代码或条件门禁,进入关键功能。一般来说可能是账号或服务端规则或ab等在当前条件下无法复现,需要使用hook强制开启。
- 禁止:修改代码。你的目标只是复现,不需要你改代码。
## Android 构建和安装
预处理已检出目标提交并应用 Security patch,尚未构建或安装 App。使用本节点的 RemoteX 构建方式构建并安装 APK,失败时排查修复后再复现。
## 推荐做法
→ 清空 frida-server,abmock,重置环境
→ 启动 frida-server,启动app
→ 注入脚本
→ 执行操作步骤
→ 观察 Hook
→ 通过脚本强制打开关键门禁
→ 观察 Hook
→ 修改脚本
→ 再次复现
→ 汇总证据
## 输入文件
先读取并以这些文件为准:
- /fixture/workspace/task-1/review_spec.md 原始的复现需求
- /fixture/workspace/task-1/machine_readable_repro_spec.json 人工复现的过程信息(仅参考价值)
- /fixture/workspace/task-1/TestSpec.md 你需要的复现路径
## Android 运行注意事项
- 如果你的复现需要多个设备,那你可能要通过hook等手段mock。
- ANR 弹窗不等于失败,一般来说应该忽略ANR弹窗。
- 如果同一场景持续 ANR、崩溃或弹窗阻塞复现,使用 Frida hook 短路关键阻塞点。
- 除了frida hook外,鼓励使用https://github.com/JingMatrix/Vector或https://github.com/LSPosed/LSPosed进行hook,注意观察端侧使用的真实框架。链接的设备具有root。
- 任务开始前记得检查有没有安装的xp模块,如果有作用于目标App的,记得卸载。
- 任务开始前,记得重置连接设备的frida-server环境,防止缓存导致的任务失败。
- 默认目标包是 `com.zhiliaoapp.musically`(Musically/M 包); 只有复现要求里明确要求 Trill/T 包时, 才使用 `com.ss.android.ugc.trill`。
- 登录频繁、拒登或 captcha 反复触发时,尝试切换账号或登录方式,实在不行需要你记录到最终文档。
- App启动可能会出现系统录屏请求,请同意录制。
## 执行事项
- ADB 设备由你根据任务自行选择;确定后设置 `TARGET_PHONE`,本轮所有 adb 命令都显式使用 `adb -s "$TARGET_PHONE"`。
- 你的hook脚本,以及编译出的xp模块,请放到`/fixture/workspace/task-1/artifacts/` 下
- 任务执行结束后,不要卸载app!!!任务执行过程中,不要卸载App!!!
- App内置了审查程序,你卸载会导致内部数据丢失。我会自己手动回收你执行完成的数据。
## 运行规则
- 按 SecuritySpec 的「复现路径」在真机上触达目标场景。
- 按 SecuritySpec 的「搜证方法」采集证据,证据一律落到 `/fixture/workspace/task-1/evidence/`:
- 关键场景截图 / 录屏:`shot-<run>-<desc>.<ext>`
- 日志 / 抓包 / hook 输出 / 状态快照:`eviden-<run>-<desc>.<ext>`
- 临时改动、hook 脚本及输出也放进 evidence
- 证据必须多截图,少文本,少日志
- 进程约束:严禁启动无上限后台命令。不要裸用 `cmd &`、`frida ... &`、`adb logcat ... &`、`sleep N | frida ... &` 这类不可回收写法;确需后台采集时必须写入 PID、设置明确时长, 并在本轮结束前停止它。优先使用有明确退出条件的脚本/driver;Frida 长时间 attach 不要依赖交互 CLI + pipe 保活。
- 按「判定标准」核对每个验证项:
- 无法复现、未复现、部分复现、被前置条件阻塞或证据不足,都要如实写完两份报告后结束本轮。
## 交付契约
- 全部产物落在 `/fixture/workspace/task-1/` 下。
- 整个 session 产生的任何截图、设备日志、logcat、抓包 脚本或输出,统统输出到 `/fixture/workspace/task-1/evidence/`。
- 禁止散落到 /tmp、TikTok repo 根或其他目录。
- 特别地:`/fixture/workspace/task-1/logs/` 是执行框架日志目录, 设备日志和 logcat 严禁写入或重定向到 logs/。这个目录不需操作。
- **`Conditions.jsonl`** — 本批确证的执行条件。**只能由 `conditions` 工具写入**,不要手动创建或编辑这个文件
- **`EvidenceReport.md`** written **entirely in English** (only identifiers stay in English); follow the template below for STRUCTURE and section ORDER, but translate its Chinese section headings/labels into natural English:
```markdown
# 采证报告
## 总结结论
- 总体结论: 全部复现 / 部分复现 / 未复现 / 证据不足
- 环境: 平台、设备、构建、账号、region、实验配置
- 关键阻塞:
## SPEC 验证项逐项对比
| ID | 原始验证点 | 结论 | 实际观察 | 原因分析 | 证据 |
|---|---|---|---|---|---|
| V1 | ... | 已复现 / 未复现 / 部分复现 / 证据不足 | ... | ... | evidence/... |
## 主路径执行记录
...
## 分支 / fallback / diagnostics
...
## 已知边界与不可证明项
...
## 证据索引
- evidence/...
```
章节命名和顺序必须与模板一致;"SPEC 验证项逐项对比"里的每个 V<n> 对应 SecuritySpec 判定标准里的一条,一一列全,不要合并、不要遗漏。
- **`EvidenceReport.json`** 同目录再写一份**机器可读**的搜证结论(纯 JSON,UTF-8,不加 markdown 围栏),schema 如下,枚举值必须严格用大写英文常量:
```json
{
"overall": "REPRODUCED | NOT_REPRODUCED | PARTIALLY_REPRODUCED | BLOCKED | EVIDENCE_INSUFFICIENT | NOT_APPLICABLE",
"environment": {
"platform": "android | ios",
"device": "...",
"build_ref": "...",
"account": "...",
"region": "...",
"experiments": { "<key>": "<value>" }
},
"blockers": ["..."],
"spec_items": [
{
"id": "V1",
"point": "原始验证点摘要",
"verdict": "REPRODUCED | NOT_REPRODUCED | PARTIALLY_REPRODUCED | EVIDENCE_INSUFFICIENT | NOT_APPLICABLE | BLOCKED",
"observation": "实际观察",
"cause": "原因分析",
"evidence": ["evidence/xxx.png", "evidence/yyy.log"]
}
],
"main_path": "主路径执行记录的一句话摘要",
"diagnostics": ["分支 / fallback / diagnostics 关键结论"],
"known_limits": ["已知边界与不可证明项"],
"evidence_index": ["evidence/..."]
}
```
枚举值枚举清单(供参考):
- overall/verdict 中的 `REPRODUCED` = 全部复现
- `NOT_REPRODUCED` = 未复现
- `PARTIALLY_REPRODUCED` = 部分复现
- `BLOCKED` = 前置条件阻塞
- `EVIDENCE_INSUFFICIENT` = 证据不足
- `NOT_APPLICABLE` = 不适用
两份产物必须内容一致:`EvidenceReport.md` 是人读版,`EvidenceReport.json` 是同一次搜证结论的结构化对照,字段与 md 里"SPEC 验证项逐项对比"表格一一对应,你不要生成或猜测.
任务上下文:
- 标题: 剪贴板越权读取
- 平台: android
- 目标构建 commitSHA: (默认分支 develop)
- 历史尝试次数: 1
## 一些相关知识库(仅参考)
## 执行条件记录(**本批执行必做,与写 EvidenceReport.md 同等强制**)
你是**第一个真正把这件事在真机上跑通的人**。TestSpec.md 里的执行条件(账号 / 入口路径 / 实验开关 / 构建方式 / 必要操作)是评估阶段**基于工单和代码调研推断出来的,没有在真机上验证过** —— 对不上是常态,写对了也只是碰巧。
所以本批请把「**复现并取证这个安全问题真正需要哪些条件**」逐条记下来。注意记录的对象是**条件本身**,不是"spec 哪里写错了":
```
/opt/homebrew/lib/node_modules/bun/bin/bun.exe /fixture/satellite.js conditions record --kind <类型> --scope "<业务模块>" --subject "<这条讲哪件事>" --statement "<条件是什么>" \
[--spec <predicted|contradicted|absent>] [--spec-said "<spec 原本怎么写的>"] \
[--payload '<JSON>'] [--evidence <evidence/ 下的文件名>] [--assumed]
```
- `--kind`:account / entry_path / experiment / env / tricky / pitfall —— 分别对应 账号权限 / UI 路径 / 实验开关 / 设备构建 / 必要操作 / 陷阱
- `--scope`:这条在**哪个业务模块内成立**,如 "Story 相机"、"消息盒子"。**要稳定可复用**,别写成本任务专有的描述
- `--subject`:这条**讲哪一件事**,如 "特效面板入口路径"、"发布所需账号权限"。**同一件事在不同任务里必须写成同一个 subject** —— 这是系统判定新旧知识是否冲突的依据,写飘了矛盾就永远发现不了
- `--statement`:条件本身,写成**将来别人能直接照做**的形式
- `--spec`:顺手标注 spec 猜得怎么样 —— `predicted`(spec 已写对)/ `contradicted`(spec 写错了,请一并给 `--spec-said`)/ `absent`(spec 没提到)。**这一项只是给 spec 质量打分用的,不影响这条条件是否该记**
- `--payload`:能结构化就结构化,如 `'{"abid":"7712334"}'` 或 `'{"path":["我的","设置","隐私"]}'`。有结构化载荷时冲突判定精确得多
- 默认按「真机已验证」记录;**只是你从代码里读出来、没实际跑通的,加 `--assumed`**
**要记的是"下一个同模块任务也用得上"的条件**:取证路径实际怎么走 / 需要哪个账号或权限才能触达 / 要开哪个实验或 mock / 必须冷启动才能观察到 / 抓包或 hook 需要什么前置。
**不要记**:本次特有的代码改动、具体行号、一次性环境故障、你自己的操作失误。**你为了做得更好而主动做的设计取舍也不算条件** —— 那写进 EvidenceReport.md。
- **发现一条就立刻记一条**,不要攒到最后(session 末尾必漏)。
- **本轮结束前执行一次** `/opt/homebrew/lib/node_modules/bun/bin/bun.exe /fixture/satellite.js conditions done`,它会汇总本批记了什么(无论最终证据结论如何都适用)。
- **拿不准算不算就记下来** —— 漏记的代价远大于多记。
- 这是你唯一被允许的"修正 spec"的方式:**严禁直接编辑 TestSpec.md 或 TestSpec.md**。
### 卡住时可以再查一次知识库
```
/opt/homebrew/lib/node_modules/bun/bin/bun.exe /fixture/satellite.js kb recall --target security:sec-1 "<你的问题,用你会对同事说的话写>"
```
prompt 顶部那份知识是**按平台和历史有用度**选的,跟这张单子写了什么无关 —— 所以入口路径、实验开关、账号权限这类**跟当前这个页面绑定**的知识,派发时不可能命中。**在真机上卡住的那一刻直接查**:
- 例:`/opt/homebrew/lib/node_modules/bun/bin/bun.exe /fixture/satellite.js kb recall --target security:sec-1 "Story 相机特效面板从哪进"`
- 命中的条目会**接着现有编号往下排**(顶部已经到 K6,新命中就是 K7、K8…),用法和顶部那些完全一样:采纳了就在报告里标 `〔K7〕`,并计入最后的 `KB_USED:`
- **查不到就是查不到**,它会直说;这时按你自己的调研走,别把「没查到」当成「不存在」
- 与你的代码调研冲突时,**永远以代码为准**
> 为什么值得花这几秒:被打回或进入下一批执行时,你读到的 EffectiveSpec.md 会自动带上这些已确证的条件,**你不会再踩同一个坑**;它们还会沉淀进蒸馏库,让以后同模块任务的 spec 从第一版就带着真实条件。
## 其他人工补充:
## Starting builds / pulls / deploys — use `wait-for`, don't hand-roll the wrapper
**Start a long command and wait for it, in one call:**
```bash
/opt/homebrew/lib/node_modules/bun/bin/bun.exe /fixture/satellite.js wait-for --run '<the build command you were going to run, verbatim>' \
--timeout 45m --label 'iOS build r2'
```
It does all four things for you: starts it detached (decoupled from this Bash
call), captures output to a log file, waits for it to finish, and prints the
**real exit code plus the last few dozen log lines**. No `nohup`, no
`echo $? > x.exit`, no `while true; do … sleep; done`.
**Wait for something already running:**
```bash
/opt/homebrew/lib/node_modules/bun/bin/bun.exe /fixture/satellite.js wait-for --pgrep 'airb pull' --timeout 20m --label 'pull app' # a process to exit
/opt/homebrew/lib/node_modules/bun/bin/bun.exe /fixture/satellite.js wait-for --pid 12345 --tail /tmp/build.log --timeout 30m # a specific pid
/opt/homebrew/lib/node_modules/bun/bin/bun.exe /fixture/satellite.js wait-for --file ./deploy.log --match 'pull success' --timeout 20m # a log marker
/opt/homebrew/lib/node_modules/bun/bin/bun.exe /fixture/satellite.js wait-for --until 'airb cmd "test -f /tmp/r3_build.exit"' --timeout 45m # any condition
```
Exit codes: **0 = success · 1 = the command failed (real code is in the output)
· 124 = timed out**. A timeout does **not** kill a `--run` command — it prints
the pid and a ready-made resume command, because a 40-minute build should not be
thrown away over a short `--timeout`. `--help` lists every flag.
**Copy the build command from this node's skill; do not retype it from memory.**
It is long and carries nested quoting (`--other_linker_flags`); getting that
escaping wrong costs a whole 25-minute build — which has already happened. Paste
it verbatim into `--run '…'`; only the outer single quotes need escaping.
**Why this is mandatory:** every Bash call you make is a full model turn —
thinking, generation, a full cache read. Polling a 25-minute build one
`pgrep` / `date` / `du -sh` at a time burns dozens of them. Measured across a
50-task sample, **byte-identical repeat calls were 14% of all Bash calls** — the
single largest pool of wasted time in this system. `wait-for` collapses it to one.
- The Bash tool's timeout ceiling is raised to one hour, so a single
`wait-for` really can outlast a build — but the **default is still 2 minutes**,
so pass an explicit `timeout` (in ms) on the Bash call for a long wait.
- A command that runs past its timeout without one is moved to the **background**,
where it holds the whole agent process open — this has already turned a
finished evaluation into a "timed out" one. Long commands either get a
timeout or get waited on with `wait-for`.
- **Never wait on device UI state with it.** Keep `catuse apps | grep -q …`,
`adb shell dumpsys window | grep -q mCurrentFocus` and `uiautomator dump | grep -q …`
out of `--until`. A build has a definite end — not finished yet just means not yet.
**UI state has none**: it may never appear (wrong page, a dialog on top, the app
crashed), and all you get back is "didn't happen", with no idea what page you are
actually on. Use `catuse ui` or a screenshot to look, then decide.
Measured: in the first three days, 33 timeouts burned 4.4 hours — **every one of
them this pattern**, with 3–8 minute timeouts (UI scale, not build scale).
- **Wait for a screen with `/opt/homebrew/lib/node_modules/bun/bin/bun.exe /fixture/satellite.js wait-for --ui-text '<text>'`, not a fixed `sleep`.** After a
tap / cold start / navigation, `/opt/homebrew/lib/node_modules/bun/bin/bun.exe /fixture/satellite.js wait-for --ui-text 'Inbox'` (regex, case-insensitive) reads the
UI tree every second and returns as soon as the text shows up; `--ui-gone 'Loading'` waits for
a spinner to go away. Default 30s, max 3m — seconds-scale, unlike the long waits above. On a
miss it **prints the text currently on screen**, so you know which page you are on without a
screenshot. Sampled rounds spent 3–21 minutes in hard-coded `sleep`s. Sub-second settles
(`sleep 1` after a tap) are still fine.
- `wait-for` only **waits**; it never starts anything. Start the build the way
this node normally does (backgrounded, dropping an exit-code file), then
`--until` that file exists.
- If you hit 124, report "waited N minutes, did not finish" as the fact it is.
**Do not rewrite it as "verification environment blocked"** — that is the most
expensive misdiagnosis this system produces, because it looks like infra rot
and nobody re-runs it.
---
SPEC SECTION NAMES — this task's spec is written in ENGLISH. Where the instructions above refer to a spec section by its Chinese name, look for its English heading instead:
证据目标 → Evidence goal · 复现前置 → Repro preconditions · 复现路径 → Repro path · 搜证方法 → Evidence method · 判定标准 → Verdict criteria · 风险与前置缺失 → Risks & gaps · 涉及的关键文件 / 模块 / 入口 → Key files, modules & entry points · 构建基线 → Build baseline · 目标平台/设备 → Target platform / device · 前置条件 → Preconditions · 判定成功的标志 → What counts as done.
The spec's content is authoritative regardless of which name the instructions used.