メニュー
Find linux-erofs mailing-list threads on lore.kernel.org, fetch every message in a thread, summarize the discussion at per-email granularity, produce a Chinese technical report, and optionally save patch series in git-am-ready form with b4.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Cross-agent peer review workflow for code reviews. Use this skill whenever the user asks Codex to review code, a diff, commit, pull request, merge request, patch, or branch and wants another code agent such as Claude Code, Codex, or a configured CLI reviewer to independently propose findings and critically audit Codex's findings before results are returned. Also use when the user mentions peer review, a second reviewer, cross-agent review, mutual review, another agent checking findings, Claude reviewing findings, or wants review findings validated rather than accepted at face value.
Generate a technical information digest from configured web sources using deterministic local scripts for fetching, filtering, validation, rendering, and optional JSON-configured webhook delivery, while the active agent performs scoring, topic deduplication, and enrichment from staged JSON artifacts. Use when asked to create a daily/weekly technical digest, rank fetched source items, produce scored_items/enriched_items artifacts, render a Chinese or English Markdown digest, or push the rendered digest through a configured webhook from RSS, Hacker News, GitHub, Reddit, or OSS Insight sources.
Windows-specific workflow for running complex Bash or shell commands inside WSL from Windows PowerShell. Use when scripts contain `$variables`, command substitutions, loops, quotes, here-documents, or multi-line bodies that would be corrupted by PowerShell or `wsl.exe bash -lc` quoting. Provides a bundled `wslsh.ps1` bridge that writes the script to a temporary `.sh` file and executes it inside WSL.
Read an arXiv paper and generate a structured reading report in Simplified Chinese. Use when the user provides an arXiv link or asks to read, analyze, summarize, or review an arXiv paper, especially for computer systems and AI topics. Parse the PDF with layout awareness, read the paper in正文顺序, always read the system architecture figure, and inspect other figures or tables only when the正文 explicitly cites them.
Standardize a local Linux kernel test workflow around QEMU, a fully installed Alpine qcow2 guest, and a self-built kernel. Use when an agent needs to clone or update Linux sources, install Alpine into a QEMU disk image, boot a custom kernel on that installed guest under QEMU/KVM, attach GDB, or turn ad-hoc VM commands into repeatable kernel-development steps.
获取网页文章URL，生成中文摘要并保存到Notion页面。当用户提供文章链接并要求总结保存到Notion时触发。
| name | linux-erofs-thread |
| description | Find linux-erofs mailing-list threads on lore.kernel.org, fetch every message in a thread, summarize the discussion at per-email granularity, produce a Chinese technical report, and optionally save patch series in git-am-ready form with b4. |
Use this skill when the user asks to investigate a linux-erofs mailing-list discussion, summarize a lore thread, understand review history, or fetch patches from a thread for direct git application.
Accept any of these:
linux-erofs, all, or another kernel list.Default output language is Simplified Chinese unless the user asks otherwise.
scripts/fetch_lore_thread.py to download a thread mbox and extract per
message JSON/Markdown reading material.scripts/save_thread_patches.sh when patches must be saved in a form that
can be applied by git am.b4 for patch extraction. Do not reconstruct patch files manually from
quoted email text unless the user explicitly asks for a best-effort fallback.If the user gives a lore URL or Message-ID, use it directly.
If the user gives search terms:
site:lore.kernel.org/linux-erofs <terms>.lore.kernel.org results over mirrors.Run:
python3 scripts/fetch_lore_thread.py \
--url <lore-url-or-message-id> \
--output-dir <artifact-dir>
The script writes:
<slug>.mbox: raw thread mbox.<slug>.messages.json: parsed message metadata and body text.<slug>.messages.md: compact per-message reading index.Use the JSON as the canonical input for analysis. Reopen the raw mbox only when the JSON body was truncated or MIME parsing looks suspicious.
Do not collapse the thread into only a cover-letter summary. For every message, capture:
Reviewed-by, Acked-by, Tested-by, Fixes,
Reported-by, Suggested-by, Closes, and maintainer instructionsFor patch emails, summarize the commit intent and touched subsystem, but keep review comments and follow-up replies as separate messages. For replies, state what exact concern, request, objection, or resolution the reply introduced.
Use this structure unless the user asks for a different format:
# <thread subject>
## 基本信息
- lore: <canonical thread url>
- message-id: <root or selected message id>
- 时间范围: <first date> - <last date>
- 参与者: <main participants>
- 邮件数: <count>
- patch 状态: <none | single patch | series, version, count>
## 总结
<2-5 bullets about the purpose, design, and final discussion state>
## 逐封邮件梳理
| # | 时间 | 作者 | 类型 | 主题 | 要点 |
|---|------|------|------|------|------|
## 技术主线
<group the thread's technical discussion by issue, not by chronology>
## 结论和未决问题
<what was accepted, rejected, requested, or left unresolved>
## Patch 获取情况
<paths to saved git-am-ready files, validation command, and result>
Keep the report grounded in email content. If you infer a thread outcome from silence, later versions, or maintainer wording, label it as an inference.
If the user asks to fetch patches, or the report would be incomplete without the patch artifact, run:
scripts/save_thread_patches.sh <lore-url-or-message-id> <patch-output-dir>
This uses b4 am -o <patch-output-dir> and produces mbox files intended for
git am.
Validate when a target git tree is available:
git am --check <patch-output-dir>/*.mbx
If validation fails, report the exact failing command and error. Do not edit the patch payload unless the user explicitly asks for a rebased or repaired patch.