发布前内容合规预审

SKILL效率

原名 Tcms Compliance Reviewer · 标识 tcms-compliance-reviewer

效率 未标注权限 许可证不明确
安装这个技能

在发布前检查事实引用、客户脱敏、产品命名、竞品规则与格式等问题,输出预审报告和修改建议,不改动原文。

一句话选型结论

按资料字段整理

发布前内容合规预审是效率类的 Agent Skill,安装方式未标注,权限级别未标注,许可证不明确,有可直接复制的安装命令。技能本质是给模型读的说明书,效果取决于模型与客户端;装完先用一个小任务验证,再放进正式流程。

分类效率 安装方式未标注 权限级别未标注权限 许可证许可证不明确
累计安装 未标注 本周 未标注
GitHub 星标 未标注 仓库 未标注
下载量 未标注 技能市场下载数
权限级别 未标注权限 安装前请确认风险范围
许可证 许可证不明确 未说明

安装

未标注
在支持 Skills 的客户端里执行,装完问一句「你有哪些技能」验证
skillhub install user_7fbe8d41/tcms-compliance-reviewer
  1. 1
    确认客户端支持

    确认你的客户端支持 Skills(Claude、Codex、Cursor 等),并升级到较新的版本。

  2. 2
    拿到技能文件

    执行上面的命令安装;也可以把仓库里的 SKILL.md 手动放进客户端的 skills 目录。

  3. 3
    刷新技能列表

    重新打开客户端,或刷新技能列表,让新放进来的技能被加载。

  4. 4
    验证是否生效

    问一句「你有哪些技能」,确认列表里出现它,再拿一个小任务试跑。

技能档案

效率
字段来自技能仓库与公开索引,中文说明由本站标注
作者 / 维护方user_7fbe8d41仓库 未标注 · 版本 未标注
宿主客户端未标注技能可被这些客户端加载
安装方式未标注安装包 community
权限级别未标注权限安装前请确认风险范围
许可证未说明许可证不明确,商用前请自行确认条款
仓库可用性available技能仓库当前是否可访问
最近核验2026-11-06上游最后核验时间 · 数据更新 2026-11-06
正文体量8.6 KBSKILL.md 原文大小,越大说明说明书写得越细

能做什么

0 项能力
上游给出的能力概览与适用场景
检查事实引用核对客户脱敏排查AI痕迹

典型使用场景

3 类
按分类整理的常见用法,具体以 SKILL.md 为准
整理待办与日程

把零散任务收进清单,排出优先级。

批量处理重复任务

把固定动作做成模板,重复执行。

把流程写成清单

把口头流程整理成可复用的步骤。

选择建议

装之前先看这几条
按条目字段给出的判断,不替你决定用不用

值得选的地方

  • 有可直接复制的安装命令,不用自己翻文档找入口。

先想清楚的地方

  • 没有标注权限级别,实际能做什么要装好后确认。
  • 许可证不明确,商用或二次分发前先到仓库确认条款。
  • 没有做站内核验,内容以作者仓库为准。
  • 技能只是说明书,实际效果取决于模型与客户端,装完先用小任务试一遍。

更适合

  • 需要在效率场景里给模型补一套做法的人。
  • 愿意先读一遍 SKILL.md、确认它会让模型做什么的人。

不太适合

  • 对许可证要求很严、必须明确可商用的人。
  • 只想装完就自动变强、不打算看说明书的人。

同类对比

6 个同类
同分类里挑最接近的几条,字段以各自详情页为准
技能权限级别安装方式累计安装宿主客户端安装命令
发布前内容合规预审当前 未标注权限 未标注 未标注 未标注 有
Nano PDF 文档 联网访问 ClawHub 市场 424.9 万 OpenClaw 有
自动更新程序 浏览器控制 ClawHub 市场 296.4 万 OpenClaw 有
Gemini 助手 联网访问 ClawHub 市场 254.7 万 OpenClaw 有
word DOCX 文档 联网访问 ClawHub 市场 239.6 万 OpenClaw 有
Excel XLSX 前端 联网访问 ClawHub 市场 227.0 万 OpenClaw 有

SKILL.md 全文

8.6 KB
模型实际读到的说明书原文,来自技能仓库

Content Compliance Reviewer

Runs a pre-publication compliance and quality review on finished tech blogs, customer cases, product updates, or press releases. Only checks and reports — never auto-modifies the original.

When to use

  • Need to check data facts, customer redaction, product official names, competitor rules, and internal information in an article.
  • Need to confirm citation traceability and formatting before publication.
  • Need a pre-review report with an approval recommendation.

Do not use

  • Monthly performance analysis and content retrospectives — handled by content-performance-analyst.
  • Auto-modifying or rewriting the original; provide only location and fix suggestions.
  • Initial writing or channel adaptation.

Input

draft_path:
article_type: tech_blog | case_study | product_update | other
compliance_profile: path or object
  - brand_rules:
  - sensitive_terms:
  - product_public_status:
  - customer_redaction_policy:
  - competitor_policy:
review_requested_by:

The compliance profile is supplied by the project's private configuration. The generic engine uses logical fields; real paths and rule tables stay in the private layer.

Workflow

Step 1: [Deterministic] Load the draft and reference

  1. Read the article under review.
  2. Read the brand rules, sensitive-term list, product-public-status, and customer-redaction rules from the compliance profile.
  3. Look up the corresponding product section in the knowledge base as needed, to verify data provenance.

Step 2: [LLM] Item-by-item inspection

# Check PASS description
1 Customer redaction No real internal customer name; description cannot identify the customer
2 Product formal name First mention uses the official full name
3 Competitor rules No competitor company or product name
4 Data traceability Each figure can be traced to an approved source
5 Internal information No internal code names, project names or unconfirmed capabilities
6 Format and structure Word count, brand title, closing and citation table comply
7 Content quality Core message is clear, logic is consistent, no obvious AI template language

Every item gets: PASS / NEEDS_FIX / FAIL.

Step 3: [LLM] Generate pre-review report

Output:

# Pre-Review Report

## Summary

| Item | Result | Note |
|---|---|---|
| Customer redaction | PASS/NEEDS_FIX/FAIL | |
| Product name | PASS/NEEDS_FIX/FAIL | |
| Competitor | PASS/NEEDS_FIX/FAIL | |
| Data traceability | PASS/NEEDS_FIX/FAIL | |
| Internal info | PASS/NEEDS_FIX/FAIL | |
| Format | PASS/NEEDS_FIX/FAIL | |
| Content quality | PASS/NEEDS_FIX/FAIL | |

## Specific issues

- Location, problem, suggested fix for each NEEDS_FIX or FAIL.

## Citation verification

| Claim | Source | Status |
|---|---|---|

## Overall assessment

- Publishable / Minor fixes / Major revision / Rewrite recommended

## Recommended approval level

- Level and rationale.

Step 4: [Deterministic] Save

Save to: content/drafts/{original-name}-compliance-review.md

Execution summary:

## Execution Summary
- Article reviewed:
- Results: N pass / M needs-fix / K fail
- Overall: publishable | minor-fix | major-fix | rewrite
- Critical issues:

Pre-publish strengthened checks (backstop for high-frequency rework points)

Beyond the seven general checks, run mandatory checks on high-frequency rework points (corresponding to the P0/P1 items found by the claim / cross-material governance lines). These are also the pre-publish backstop for content-writer's expression red lines:

  • P0 naming consistency: the tech-blog product's external name must match verbatim the already-published authoritative materials of the same campaign (press release / official site / official account); must not add its own version number or suffix (e.g. if the external name is uniformly "X", the blog must not write "X 2.0"). Inconsistency = FAIL.
  • P0 meta-language / self-reference: sentences that break the frame — "this article…", "the press release makes it clear that…", "deserves its own section", "back to the overall narrative" — = FAIL; replace with a direct content transition.
  • P0 business jargon: four-character slogans like "多、快、好、省" = FAIL; rewrite in engineering dimensions (load coverage / execution efficiency / operations experience / resource efficiency).
  • P1 absolutist phrasing: scan for "naturally connected / seamless / inevitable / certain / zero" etc.; require a bounded qualifier instead.
  • P1 out-of-scope scenarios: landing industries/scenarios without a knowledge-base or published-article source must be marked [needs confirmation] or removed; do not list scenarios from memory (e.g. an industry/scenario needs internal-source backing).
  • P1 faithful transcription vs quantified claims: infrastructure capabilities backed by an architecture diagram may be stated; quantified speedup multiples without an official figure = FAIL.

> When publishing externally and multiple materials on the same topic already exist, after a PASS pre-review it is recommended to also run the three-piece governance line — claim-to-source-auditor + cross-material-consistency-auditor + tech-content-review-panel — before finalizing.

Hard Rules

  1. Never modify the original draft. Only report issues and suggestions.
  2. Data checks must reference an approved source. Missing sources receive a fail.
  3. Any real customer name hit is an automatic fail.
  4. Employer or unrelated brand names must not appear in outward-facing content unless the piece is officially co-branded; a hit is a fail even when the sentence is otherwise correct.
  5. Unconfirmed product capabilities receive a needs-fix.
  6. Competitor named references must be flagged.
  7. Internal code names or project names are an automatic fail.
  8. Human confirmation required before marking the review complete and handing off to adaptation.

Failure Handling

Scenario Action
Draft not found List available drafts for the user
Compliance profile missing Warn and skip brand/sensitive-term checks; flag as incomplete
Knowledge-base section unavailable Mark all data items as unverifiable
Draft has no citation table Mark as recommended
Competitor, customer or product policy changed Request updated profile before proceeding

Output Format

content/drafts/{original-name}-compliance-review.md

Verification

  • [ ] All seven inspection items evaluated.
  • [ ] Every data claim checked against approved source.
  • [ ] Every real customer name flagged.
  • [ ] Report saved without modifying the original draft.
  • [ ] Execution summary includes next action.

中文摘要

Content Compliance Reviewer 是发布前合规与质量预审 Skill:检查事实引用、客户脱敏、产品口径、竞品规则、内部信息、格式与 AI 痕迹,只报告不修改原文。除七项通用检查外,对高发返工点做强制兜底(P0 命名一致性/元语言/商务腔,P1 绝对化/超范围场景/量化断言),与 content-writer 的表达红线前后呼应。多份同主题物料对外发布前,建议再跑 claim-to-source + cross-material + tech-content-review 三件套治理线。

常见问题

5 问
答案由本站依据条目字段整理
发布前内容合规预审怎么安装?

在支持 Skills 的客户端里执行:skillhub install user_7fbe8d41/tcms-compliance-reviewer。安装方式为未标注。装完刷新技能列表或重开客户端。

发布前内容合规预审支持哪些客户端?

资料没有标注宿主客户端,但只要客户端支持 Skills(Claude、Codex、Cursor 等),一般都能加载。

发布前内容合规预审权限高吗,安全吗?

权限级别为未标注权限。安装前先看它会读取或写入哪些内容。

发布前内容合规预审收费吗?

本站不代收费用,也没有标注具体价格。是否收费取决于技能本身和所用的模型或客户端,正式使用前按作者说明确认。

发布前内容合规预审用什么许可证?

许可证标注为 未说明(许可证不明确)。商用或二次分发前请到仓库确认条款原文。

热门专题

按场景挑技能
全部专题 →

技能内容以作者仓库为准;分类、权限级别与许可证状态的中文说明为中文整理,最近核验 2026-11-06。