Token导航 LogoToken导航

Agent Skill · 开发

canonical-format-checker(规范格式检查器)

canonical-format-checker 用于处理 GitHub 仓库、Issue、Pull Request 和代码协作信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要围绕仓库状态、代码变更或协作事项进行整理时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

canonical-format-checker 是开发类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为只读;许可证信息还不明确,第三方扫描有未通过项,按安装量在开发类的 21,746 个 Skill 里排第 12,914。

总安装 1,082周安装 46GitHub Stars 158下载 379核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
unknown

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:canonical-format-checker(规范格式检查器)
来源仓库:https://github.com/panaversity/agentfactory
仓库路径:skills/canonical-format-checker
安装命令:
npx skills add https://github.com/panaversity/agentfactory --skill canonical-format-checker
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/panaversity/agentfactory --skill canonical-format-checker

这个 Skill 是做什么的

开发类 Skill 面向写代码、改代码和排查问题,一般会读写仓库文件,部分还会执行构建或测试命令。

canonical-format-checker 防止内容格式偏离权威源,确保技能文档与已有规范一致。适用于教学类任务,在Codex、Claude中校验模式是否符合平台既定标准。自动触发检查,在实现前比对Chapter 5等基准格式,减少知识冲突。需配合ADR-115等规范文档使用,侧重结构而非语义变更。

适用宿主包括 Codex、Claude、Cursor、Gemini CLI,接入前应确认版本、权限和运行环境要求。

来源仓库内路径:skills/canonical-format-checker

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 Codex、Claude、Cursor、Gemini CLI;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 npx skills add https://github.com/panaversity/agentfactory --skill canonical-for…;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它标注为只读,先用只读任务试一次,确认行为符合预期再扩大使用范围。

常见问题

6 条

canonical-format-checker 怎么安装?

在终端执行 npx skills add https://github.com/panaversity/agentfactory --skill canonical-format-checke…。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

canonical-format-checker 支持哪些 AI 工具?

公开资料标注支持 Codex、Claude、Cursor、Gemini CLI。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

canonical-format-checker 需要什么权限?

权限级别是只读。该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

canonical-format-checker 是免费的吗?

公开资料没有标注明确许可证,收费和商用限制都无法确认。正式使用前请到来源仓库核对授权说明。

canonical-format-checker 用的人多吗?

公开数据是累计安装 1,082 次,近一周安装 46 次,下载 379 次,GitHub 158 星。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

canonical-format-checker 安全吗?

第三方扫描有未通过项,风险较高。除非你确认过源码,否则不建议直接安装到常用环境。

值得选的地方和要先想清楚的地方

值得选的地方

  • 同时适配 4 种工具,换客户端不用重新找替代品。
  • 标注为只读权限,试用阶段不容易误改本地文件。

先想清楚的地方

  • 许可证没有明确标注,能不能商用、能不能再分发都不确定。
  • 第三方扫描存在提醒项或未通过项,需要人工逐条确认。
  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 Codex,想直接扩展它能力的人。
  • 需要处理开发类任务、又不想每次都手写提示词的人。
  • 只想先试试效果、不希望它改动本地文件的人。

不太适合

  • 需要明确授权才能用到公司项目里的场景。
  • 需要厂商级技术支持和 SLA 的生产系统。

开发类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
skill-vetter

技能审查员 · OpenClaw

8,779,018986操作浏览器MIT-0通过
ontology

本体论 · OpenClaw

6,636,960576只读MIT-0通过
entra-app-registration

内部应用程序注册 · Codex、Claude

6,631,872764接触敏感数据未明确通过
entra-app-registration

内部应用程序注册 · Codex、Claude

3,757,680201接触敏感数据未明确通过
nano-banana-pro

纳米香蕉专业版 · OpenClaw

3,526,898364接触敏感数据MIT-0有提醒
canonical-format-checker

规范格式检查器 · 本页

1,082158只读未明确未通过

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

Canonical Format Checker

Overview

Prevent format drift by verifying content follows authoritative canonical sources. When teaching a pattern that exists elsewhere in the platform (skills, subagents, ADRs, etc.), this skill ensures the taught format matches the canonical source.

Why this matters: The Chapter 14 format drift failure occurred because the lesson taught skill format differently from Chapter 5 (canonical source). Students learned incorrect patterns that contradicted earlier chapters.

When to Use This Skill

Automatic Triggers (check before implementation):

  • Lesson teaches how to create skills → Check .claude/skills/ structure
  • Lesson teaches how to write subagents → Check .claude/agents/ structure
  • Lesson teaches ADR format → Check specs/*/adrs/ structure
  • Lesson teaches PHR format → Check history/prompts/ structure
  • Lesson teaches specification format → Check specs/*/spec.md structure

Manual Triggers (user requests):

  • "Check canonical format for skills"
  • "Verify format consistency"
  • "Does this match the canonical source?"

Canonical Source Lookup Table

Pattern Being TaughtCanonical Source LocationKey Format Elements
Skills.claude/skills/<name>/SKILL.mdDirectory structure, YAML frontmatter with name, description
Subagents.claude/agents/<name>.mdYAML frontmatter with name, description, model, skills
ADRsspecs/<feature>/adrs/adr-*.mdNumbered files, standard ADR structure
PHRshistory/prompts/<stage>/Template from .specify/templates/phr-template.prompt.md
Specificationsspecs/<feature>/spec.mdSections: Overview, User Stories, FRs, SCs, Evals
Commands.claude/commands/*.mdYAML frontmatter, step-by-step phases

Workflow

Step 1: Identify What Pattern Is Being Taught

Ask yourself:

  • Does this lesson teach how to CREATE any platform pattern?
  • Is there an existing canonical source for this pattern?
Trigger Examples:
- "Lesson teaches students to write their own skills" → Skills pattern
- "Chapter covers custom commands" → Commands pattern
- "Module explains agent creation" → Subagents pattern

Step 2: Locate and Read Canonical Source

MANDATORY: Read the canonical source BEFORE writing or reviewing content.

# For Skills - read actual skill structure
ls .claude/skills/*/
cat .claude/skills/session-intelligence-harvester/SKILL.md

# For Agents - read actual agent structure
ls .claude/agents/
cat .claude/agents/content-implementer.md

# For Commands - read actual command structure
ls .claude/commands/
head -50 .claude/commands/sp.specify.md

Step 3: Extract Format Requirements

Document the required format elements from canonical source:

## Canonical Format: Skills

**Directory Structure**:

.claude/skills/ └── / # Directory, NOT flat file └── SKILL.md # SKILL.md file (uppercase)

**YAML Frontmatter** (REQUIRED):

name: "<skill-name>" description: "This skill should be used when... Use when..."



**Content Structure**:

1. H1 title matching skill name
2. Overview section explaining purpose
3. When to Use section with triggers
4. Workflow section with steps

Step 4: Compare Content Against Canonical

Check the lesson/content for format consistency:

## Format Comparison Checklist

**Skill Format Check**:
- [ ] Shows directory structure (not flat file)
- [ ] Uses SKILL.md filename (not skill.md or index.md)
- [ ] Includes YAML frontmatter with `name` and `description`
- [ ] `description` starts with "This skill should be used when..."

**Agent Format Check**:
- [ ] Shows single .md file in agents directory
- [ ] Includes YAML frontmatter with `name`, `description`, `model`
- [ ] Includes `skills` array (can be empty)
- [ ] Description explains when to invoke

**Command Format Check**:
- [ ] Shows .md file in commands directory
- [ ] Has numbered phases with clear gates
- [ ] Includes enforcement checks

Step 5: Report Drift or Confirm Compliance

If Format Drift Detected:

## Format Drift Detected

**Pattern**: Skills
**Location**: Lesson 7, Section "Creating Your First Skill"

**Issue**: Shows flat file `.claude/skills/my-skill.md`
**Canonical**: Directory structure `.claude/skills/my-skill/SKILL.md`

**Specific Fixes Required**:
1. Line 45: Change `.claude/skills/my-skill.md` to `.claude/skills/my-skill/SKILL.md`
2. Line 52: Update example to show directory creation with `mkdir`
3. Line 60: Add YAML frontmatter example with required fields

**Cross-Reference**: Chapter 5, Lesson 7 (agent-skills.md) is canonical source

If Format Compliant:

## Format Verified

**Pattern**: Skills
**Canonical Source**: `.claude/skills/session-intelligence-harvester/SKILL.md`
**Content Location**: Chapter 14, Lesson 5

**Verification**:
- [x] Directory structure matches canonical
- [x] YAML frontmatter format correct
- [x] Required fields present (name, description)
- [x] Description follows "This skill should be used when..." pattern

**Status**: COMPLIANT - No drift detected

Common Format Drift Patterns

Drift Pattern 1: Flat Files vs Directories

Wrong: .claude/skills/my-skill.md Right: .claude/skills/my-skill/SKILL.md

Drift Pattern 2: Missing YAML Frontmatter

Wrong: Markdown file with no frontmatter Right: YAML frontmatter with required fields

Drift Pattern 3: Wrong Filename

Wrong: skill.md, index.md, README.md Right: SKILL.md (uppercase)

Drift Pattern 4: Missing Required Fields

Wrong: Only name in frontmatter Right: Both name and description (description starts with "This skill should be used when...")

Drift Pattern 5: Inventing New Format

Wrong: Creating custom format not matching canonical Right: Following exact structure from canonical source

Integration Points

Use With:

  • chapter-planner - Check during lesson planning phase
  • content-implementer - Verify before implementation
  • validation-auditor - Include in validation checklist

Invoke When:

  • Planning lessons that teach platform patterns
  • Reviewing content that describes file structures
  • Validating documentation about Claude Code features

Self-Monitoring

Before approving content that teaches patterns:

  • Identified what pattern is being taught
  • Located and READ the canonical source
  • Extracted format requirements from canonical
  • Compared content against canonical requirements
  • Detected drift OR confirmed compliance
  • Provided specific fix instructions if drift found

Success Criteria

You succeed when:

  • Format drift detected BEFORE publication
  • Specific canonical source identified and referenced
  • Exact format requirements documented
  • Content matches canonical source exactly

You fail when:

  • Content teaches incorrect format that contradicts canonical
  • Canonical source not checked before implementation
  • Vague feedback ("fix the format") instead of specific corrections
  • New patterns invented instead of following canonical

平台分布

Codex

34.71%
按下载量换算132

Claude

30.07%
按下载量换算114

Cursor

18.24%
按下载量换算69

Gemini CLI

9.29%
按下载量换算35

安全审计

Gen Agent Trust Hub

未通过

Socket

通过

Snyk

通过

权限和风险

只读

该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。来源安全扫描存在提醒或未通过项,安装前请自行复核。目前只有一个公开来源,建议同时核对源仓库。

继续浏览同类 Skills