From wewrite
Reviews WeWrite editorial drafts against task briefs, fact-checks claims and personal materials, scores on five editorial dimensions, and rewrites until passing. Responds to "检查这篇文章".
How this skill is triggered — by the user, by Claude, or both
Slash command
/wewrite:wewrite-reviewThis skill is limited to the following tools:
The summary Claude sees in its skill listing — used to decide when to auto-load this skill
用户指定文章时检查该文件;否则运行 `wewrite run show`,优先读取 `artifacts.draft`,并读取
用户指定文章时检查该文件;否则运行 wewrite run show,优先读取 artifacts.draft,并读取
artifacts.brief、artifacts.claims 和 artifacts.sources。旧任务没有初稿产物时才回退
artifacts.article。管道内先运行 wewrite run step review in_progress。
完整读取:
读取: {skill_dir}/../wewrite-write/references/article-brief.md
读取: {skill_dir}/../wewrite-write/references/editorial-quality.md
读取: {skill_dir}/references/seo-rules.md
用户只说“检查一下”且给了外部文件时,只给报告;主流程、任务内初稿或用户明确说“优化”时, 必须直接完成必要修改。
检查文章是否真的回答 audience.question,核心判断是否与 thesis 一致,每节是否推动指定
claim,结尾是否交付 goal.takeaway/action。无关段落删除,不用漂亮结构掩盖答非所问。
运行 wewrite sources list --json,把正文中的具体数字、日期、引述、研究结论和时效性事实
逐项对到 claims.yaml 与原始来源。来源不支持时优先补查;查不到就删除、缩小或改成明确的
推断。不能用模型记忆补洞。
检查所有第一人称事件、朋友同事、采访、对话、时间地点和感官细节。它们必须来自本次任务 明确记录的用户材料;否则属于阻断问题,直接删除或改成非亲历论述。
按“准确、观点、有用、合声、好读”各评 1-5 分,并列出阻断问题和最多 5 个主要问题:
pass:平均分至少 4、单项不低于 3、没有阻断问题。revise:能在现有材料内修正。直接修改后重新执行 1-3,不得只写建议。needs_input:仅限用户明确要求个人故事而材料不足且无法安全换框架。最多两轮。第二轮仍有问题时,删掉不可靠内容、缩小承诺,生成能通过的可靠版本;不得给未
通过的文章贴上“可交付”。通过后把最终正文写入 artifacts.article。
生成一个主标题、两个准确的备选标题、40 字内摘要和 3-5 个标签。关键词自然出现,不按密度 硬塞,也不为打开率虚构数字或承诺。审稿不调用配图。
运行 wewrite score {article_path} --json。该分数只帮助定位套话、句式过齐和段落节奏风险,
不设机械及格线,也不为提分反复重写。
先把编辑判断写为任务目录内临时 assessment.yaml:
decision: pass
pass_number: 1
dimensions:
accuracy: 4
viewpoint: 4
usefulness: 4
voice: 4
readability: 4
blockers: []
major_issues: []
notes: ""
再生成 artifacts.review_report:
wewrite content-eval --draft {draft_path} --final {article_path} \
--assessment {assessment_path} --output {review_report_path} --json
只有报告里的 publishable=true 才能完成审稿。更新任务并标记完成:
wewrite run update --patch '{"editorial":{"decision":"pass","pass_number":1,"publishable":true},"seo":{"title":"...","alt_titles":[],"digest":"...","tags":[],"quality_score":0},"provenance":{"verified_sources":0,"unverified_sources":0}}'
wewrite run step review completed
自检报告用自然语言,最多列 5 个按影响排序的问题,并说明已经怎样修正。没有硬伤就直说正文 可以交付;如有需要可再单独配图或排版,不输出大段分数表。
npx claudepluginhub imraywang/wewriteRuns a six-section quality checklist (content, structure, clarity, style, polish, final tests) before sharing or publishing writing. Use for final pre-publish reviews.
Structurally edits and improves article drafts by reordering sections, tightening arguments, and improving clarity. Interactive section-by-section process with user confirmation.
Detects AI voice patterns, measures readability, audits consistency, and validates SEO requirements on finished drafts. Use for final polish before publication.