用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/microwind/ai-skills --skill markdown命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
基于 SOC 职业分类
正在显示 SKILL.md
| name | Markdown验证器 |
| description | 当验证Markdown文件时，检查文档，查找损坏链接，改进文档。发布前验证Markdown。 |
| license | MIT |
文档会静默损坏。损坏的链接、无效语法和缺失文件使文档不可用。发布前必须验证。
核心原则: 文档即代码。需要验证。好的文档管理应该完整、准确、可访问、易维护。坏的文档管理会导致信息混乱、用户体验差、维护困难。
始终:
触发短语:
❌ 损坏链接
❌ 语法错误
❌ 内容问题
❌ 格式问题
#!/usr/bin/env python3
import re
import os
import requests
import yaml
from pathlib import Path
from typing import Dict, List, Optional, Tuple, Set
from dataclasses import dataclass
from enum import Enum
from urllib.parse import urljoin, urlparse
import time
class ValidationLevel(Enum):
"""验证级别"""
ERROR = "error"
WARNING = "warning"
INFO = "info"
@dataclass
class ValidationIssue:
"""验证问题"""
level: ValidationLevel
line_number: int
column: int
message: str
suggestion: Optional[str] = None
rule: Optional[str] = None
@dataclass
class LinkInfo:
"""链接信息"""
text: str
url: str
line_number: int
column: int
link_type: str # internal, external, image, anchor
:
():
.file_path = Path(file_path)
.base_path = Path(base_path) base_path .file_path.parent
.content =
.lines = []
.issues: [ValidationIssue] = []
.links: [LinkInfo] = []
.headers: [[, , ]] = []
.validation_rules = {
: ._validate_heading_hierarchy,
: ._validate_link_format,
: ._validate_code_blocks,
: ._validate_list_structure,
: ._validate_table_format,
: ._validate_image_references
}
() -> [ValidationIssue]:
.issues.clear()
._read_file():
.issues
._parse_content()
rule_name, rule_func .validation_rules.items():
:
rule_func()
Exception e:
.issues.append(ValidationIssue(
level=ValidationLevel.ERROR,
line_number=,
column=,
message=,
rule=rule_name
))
._check_links()
.issues
() -> :
:
(.file_path, , encoding=) f:
.content = f.read()
.lines = .content.splitlines()
FileNotFoundError:
.issues.append(ValidationIssue(
level=ValidationLevel.ERROR,
line_number=,
column=,
message=
))
UnicodeDecodeError:
.issues.append(ValidationIssue(
level=ValidationLevel.ERROR,
line_number=,
column=,
message=
))
():
line_num, line (.lines, ):
._parse_headers(line, line_num)
._parse_links(line, line_num)
():
header_match = re.(, line)
header_match:
level = (header_match.group())
text = header_match.group().strip()
.headers.append((text, level, line_num))
():
link_pattern =
re.finditer(link_pattern, line):
text = .group()
url = .group()
column = .start() +
link_type = ._determine_link_type(url)
.links.append(LinkInfo(
text=text,
url=url,
line_number=line_num,
column=column,
link_type=link_type
))
image_pattern =
re.finditer(image_pattern, line):
alt = .group()
url = .group()
column = .start() +
.links.append(LinkInfo(
text=alt,
url=url,
line_number=line_num,
column=column,
link_type=
))
() -> :
url.startswith():
url.startswith() url.startswith():
url.startswith():
url.startswith() url url:
:
():
prev_level =
text, level, line_num .headers:
prev_level > :
level > prev_level + :
.issues.append(ValidationIssue(
level=ValidationLevel.WARNING,
line_number=line_num,
column=,
message=,
suggestion=,
rule=
))
prev_level = level
():
line_num, line (.lines, ):
open_brackets = line.count()
close_brackets = line.count()
open_parens = line.count()
close_parens = line.count()
open_brackets != close_brackets:
.issues.append(ValidationIssue(
level=ValidationLevel.ERROR,
line_number=line_num,
column=,
message=,
suggestion=,
rule=
))
open_parens != close_parens:
.issues.append(ValidationIssue(
level=ValidationLevel.ERROR,
line_number=line_num,
column=,
message=,
suggestion=,
rule=
))
empty_link_pattern =
re.search(empty_link_pattern, line):
.issues.append(ValidationIssue(
level=ValidationLevel.WARNING,
line_number=line_num,
column=,
message=,
suggestion=,
rule=
))
():
in_code_block =
code_block_start =
code_fence =
line_num, line (.lines, ):
fence_match = re.(, line)
fence_match:
current_fence = fence_match.group()
language = fence_match.group()
in_code_block:
in_code_block =
code_block_start = line_num
code_fence = current_fence
current_fence == code_fence:
in_code_block =
code_fence =
:
.issues.append(ValidationIssue(
level=ValidationLevel.ERROR,
line_number=line_num,
column=,
message=,
suggestion=,
rule=
))
in_code_block:
.issues.append(ValidationIssue(
level=ValidationLevel.ERROR,
line_number=code_block_start,
column=,
message=,
suggestion=,
rule=
))
():
list_stack = []
line_num, line (.lines, ):
stripped = line.lstrip()
list_match = re.(, stripped)
list_match:
indent = (list_match.group())
marker = list_match.group()
list_stack:
last_indent, last_marker = list_stack[-]
(marker.isdigit() last_marker.isdigit()) \
( marker.isdigit() last_marker.isdigit()):
.issues.append(ValidationIssue(
level=ValidationLevel.WARNING,
line_number=line_num,
column=indent + ,
message=,
suggestion=,
rule=
))
list_stack.append((indent, marker))
stripped line.startswith() line.startswith():
list_stack.clear()
():
in_table =
table_start =
column_count =
line_num, line (.lines, ):
line:
in_table:
in_table =
table_start = line_num
column_count = line.count() -
:
current_columns = line.count() -
current_columns != column_count:
.issues.append(ValidationIssue(
level=ValidationLevel.ERROR,
line_number=line_num,
column=,
message=,
suggestion=,
rule=
))
:
in_table:
in_table =
column_count =
():
link .links:
link.link_type == :
link.url.startswith() link.url.startswith():
image_path = .base_path / link.url
image_path.exists():
.issues.append(ValidationIssue(
level=ValidationLevel.ERROR,
line_number=link.line_number,
column=link.column,
message=,
suggestion=,
rule=
))
():
link .links:
link.link_type == :
._check_internal_link(link)
link.link_type == :
._check_external_link(link)
link.link_type == :
._check_anchor_link(link)
():
link.url.startswith():
target_path = .base_path / link.url[:]
link.url.startswith():
target_path = .base_path / link.url
:
target_path = .base_path / link.url
target_path.exists():
.issues.append(ValidationIssue(
level=ValidationLevel.ERROR,
line_number=link.line_number,
column=link.column,
message=,
suggestion=,
rule=
))
():
:
response = requests.head(link.url, timeout=, allow_redirects=)
response.status_code >= :
.issues.append(ValidationIssue(
level=ValidationLevel.WARNING,
line_number=link.line_number,
column=link.column,
message=,
suggestion=,
rule=
))
requests.RequestException:
.issues.append(ValidationIssue(
level=ValidationLevel.WARNING,
line_number=link.line_number,
column=link.column,
message=,
suggestion=,
rule=
))
():
anchor = link.url[:]
found =
text, level, line_num .headers:
anchor_id = text.lower()
anchor_id = re.sub(, , anchor_id)
anchor_id = re.sub(, , anchor_id)
anchor_id == anchor:
found =
found:
.issues.append(ValidationIssue(
level=ValidationLevel.WARNING,
line_number=link.line_number,
column=link.column,
message=,
suggestion=,
rule=
))
() -> :
.validate()
error_count = ([i i .issues i.level == ValidationLevel.ERROR])
warning_count = ([i i .issues i.level == ValidationLevel.WARNING])
info_count = ([i i .issues i.level == ValidationLevel.INFO])
issues_by_rule = {}
issue .issues:
rule = issue.rule
rule issues_by_rule:
issues_by_rule[rule] = []
issues_by_rule[rule].append(issue)
{
: (.file_path),
: {
: (.issues),
: error_count,
: warning_count,
: info_count,
: (.links),
: (.headers)
},
: [
{
: issue.level.value,
: issue.line_number,
: issue.column,
: issue.message,
: issue.suggestion,
: issue.rule
}
issue .issues
],
: {
rule: [
{
: issue.line_number,
: issue.column,
: issue.message,
: issue.suggestion
}
issue issues
]
rule, issues issues_by_rule.items()
},
: [
{
: link.text,
: link.url,
: link.line_number,
: link.link_type
}
link .links
],
: [
{
: text,
: level,
: line_num
}
text, level, line_num .headers
]
}
__name__ == :
argparse
parser = argparse.ArgumentParser(description=)
parser.add_argument(, =)
parser.add_argument(, =)
parser.add_argument(, =)
parser.add_argument(, choices=[, ], default=, =)
parser.add_argument(, action=, =)
args = parser.parse_args()
validator = MarkdownValidator(args.file, args.base_path)
:
report = validator.generate_report()
args. == :
args.output:
(args.output, , encoding=) f:
json.dump(report, f, indent=, ensure_ascii=)
()
:
(json.dumps(report, indent=, ensure_ascii=))
:
( * )
()
( * )
()
summary = report[]
()
()
()
()
()
()
()
level_names = {
: ,
: ,
:
}
level [, , ]:
issues = [i i report[] i[] == level]
issues:
()
issue issues:
()
issue[]:
()
()
summary[] == :
()
:
()
Exception e:
()
exit()
#!/bin/bash
# markdown-batch-checker.sh - 批量Markdown检查工具
set -e
# 配置
SCAN_DIR=${1:-"."}
OUTPUT_DIR=${2:-"markdown-reports"}
CHECK_EXTERNAL=${3:-false}
PARALLEL_JOBS=${4:-4}
LOG_FILE="markdown_batch_check.log"
# 颜色输出
RED='\033[0;31m'
GREEN='\033[0;32m'
YELLOW='\033[1;33m'
BLUE='\033[0;34m'
NC='\033[0m'
log_info() {
echo -e "${GREEN}[INFO]${NC} $1" | tee -a "$LOG_FILE"
}
log_warn() {
echo -e "${YELLOW}[WARN]${NC} $1" | tee -a "$LOG_FILE"
}
log_error() {
echo -e "${RED}[ERROR]${NC} $1" | tee -a "$LOG_FILE"
}
log_step() {
echo -e "${BLUE}[STEP]${NC} $1" | tee -a "$LOG_FILE"
}
() {
[ ! -d ];
-p
log_info
}
() {
log_step
markdown_files=()
IFS= -r -d file;
markdown_files+=()
< <(find -name - f -print0)
log_info
>
}
() {
file_path=
file_basename=$( .md)
report_file=
log_info
-v python3 &> /dev/null;
python3 markdown_validator.py --output --format json 2>>
log_error
1
[ $? -eq 0 ];
log_info
0
log_error
1
}
() {
total_files=
checked_files=0
failed_files=0
log_step
IFS= -r file_path;
(( $(jobs -r | wc -l) >= PARALLEL_JOBS ));
-n
{
check_single_file ;
>>
>>
} &
((checked_files++))
(( checked_files % == ));
log_info
<
[ -f ];
failed_files=$(grep -c || )
successful_files=$(grep -c || )
failed_files=0
successful_files=0
log_info
}
() {
log_step
summary_file=
> <<
total_files=$( -l < )
>>
[ -f ];
failed_files=$(grep -c || )
successful_files=$(grep -c || )
>>
>>
>>
total_errors=0
total_warnings=0
total_issues=0
report_file /*_report.json;
[ -f ];
-v jq &> /dev/null;
errors=$(jq -r 2>/dev/null || )
warnings=$(jq -r 2>/dev/null || )
issues=$(jq -r 2>/dev/null || )
total_errors=$((total_errors + errors))
total_warnings=$((total_warnings + warnings))
total_issues=$((total_issues + issues))
>>
>>
>>
>>
>>
[ -f ];
failed_list=$(grep | -d: -f2-)
[ -n ];
>>
| sed >>
>>
>>
>>
>>
temp_file=
>
report_file /*_report.json;
[ -f ];
file_name=$( _report.json).md
-v jq &> /dev/null;
errors=$(jq -r 2>/dev/null || )
warnings=$(jq -r 2>/dev/null || )
issues=$(jq -r 2>/dev/null || )
>>
[ -f ];
-t -k4 -nr | -10 | IFS= -r file errors warnings issues;
>>
>>
>>
>>
>>
>>
>>
log_info
}
() {
log_info
-f
-f
}
() {
log_info
log_info
log_info
create_output_dir
total_files=$(find_markdown_files)
[ -eq 0 ];
log_warn
0
failed_count=0
check_files_parallel
failed_count=$?
generate_summary_report
cleanup
[ -eq 0 ];
log_info
0
log_warn
1
}
() {
}
[ = ] || [ = ];
show_help
0
main
#!/usr/bin/env python3
import re
import json
from pathlib import Path
from typing import Dict, List, Optional, Tuple
from dataclasses import dataclass
from collections import Counter
@dataclass
class QualityMetric:
"""质量指标"""
name: str
score: float
max_score: float
description: str
issues: List[str]
class DocumentQualityScorer:
"""文档质量评分器"""
def __init__(self, file_path: str):
self.file_path = Path(file_path)
self.content = ""
self.lines = []
self.metrics: List[QualityMetric] = []
# 评分权重
self.weights = {
'structure': 0.25,
'content': 0.30,
'links': 0.20,
'formatting': 0.15,
'readability': 0.10
}
() -> :
._read_file():
{: }
._analyze_structure()
._analyze_content()
._analyze_links()
._analyze_formatting()
._analyze_readability()
total_score = ._calculate_total_score()
{
: (.file_path),
: total_score,
: ._get_grade(total_score),
: [
{
: metric.name,
: metric.score,
: metric.max_score,
: (metric.score / metric.max_score) * ,
: metric.description,
: metric.issues
}
metric .metrics
],
: ._get_recommendations()
}
() -> :
:
(.file_path, , encoding=) f:
.content = f.read()
.lines = .content.splitlines()
:
():
issues = []
score =
max_score =
headers = ._extract_headers()
headers:
issues.append()
:
has_h1 = (level == _, level, _ headers)
has_h1:
issues.append()
:
score +=
prev_level =
hierarchy_issues =
_, level, _ headers:
prev_level > level > prev_level + :
hierarchy_issues +=
prev_level = level
hierarchy_issues == :
score +=
hierarchy_issues <= :
score +=
issues.append()
:
issues.append()
._has_toc():
score +=
:
issues.append()
(headers) >= :
score +=
:
issues.append()
word_count = (.content.split())
word_count >= :
score +=
:
issues.append()
.metrics.append(QualityMetric(
name=,
score=score,
max_score=max_score,
description=,
issues=issues
))
():
issues = []
score =
max_score =
code_blocks = ._count_code_blocks()
code_blocks > :
score +=
:
issues.append()
images = ._count_images()
images > :
score +=
:
issues.append()
tables = ._count_tables()
tables > :
score +=
:
issues.append()
lists = ._count_lists()
lists > :
score +=
:
issues.append()
._has_todo_comments():
score +=
:
issues.append()
desc_lines = ( line .lines (line.strip()) > )
total_lines = ([line line .lines line.strip()])
total_lines > desc_lines / total_lines > :
score +=
:
issues.append()
.metrics.append(QualityMetric(
name=,
score=score,
max_score=max_score,
description=,
issues=issues
))
():
issues = []
score =
max_score =
links = ._extract_links()
links:
issues.append()
.metrics.append(QualityMetric(
name=,
score=,
max_score=max_score,
description=,
issues=issues
))
(links) >= :
score +=
(links) >= :
score +=
:
issues.append()
link_types = (link.link_type link links)
(link_types) >= :
score +=
(link_types) >= :
score +=
:
issues.append()
anchor_links = [link link links link.link_type == ]
anchor_links:
score +=
:
issues.append()
external_links = [link link links link.link_type == ]
external_links:
score +=
:
issues.append()
.metrics.append(QualityMetric(
name=,
score=score,
max_score=max_score,
description=,
issues=issues
))
():
issues = []
score =
max_score =
._has_proper_code_blocks():
score +=
:
issues.append()
._has_proper_lists():
score +=
:
issues.append()
._has_proper_tables():
score +=
:
issues.append()
._has_proper_spacing():
score +=
:
issues.append()
.metrics.append(QualityMetric(
name=,
score=score,
max_score=max_score,
description=,
issues=issues
))
():
issues = []
score =
max_score =
sentences = re.split(, .content)
sentences = [s.strip() s sentences s.strip()]
sentences:
avg_sentence_length = ((s.split()) s sentences) / (sentences)
<= avg_sentence_length <= :
score +=
< avg_sentence_length <= :
score +=
issues.append()
:
issues.append()
paragraphs = [p.strip() p .content.split() p.strip()]
long_paragraphs = ( p paragraphs (p.split()) > )
long_paragraphs == :
score +=
long_paragraphs <= :
score +=
:
issues.append()
complex_words = ._count_complex_words()
total_words = (.content.split())
total_words > :
complex_ratio = complex_words / total_words
complex_ratio <= :
score +=
complex_ratio <= :
score +=
:
issues.append()
._has_excessive_repetition():
score +=
:
issues.append()
.metrics.append(QualityMetric(
name=,
score=score,
max_score=max_score,
description=,
issues=issues
))
() -> [[, , ]]:
headers = []
line_num, line (.lines, ):
= re.(, line)
:
level = (.group())
text = .group().strip()
headers.append((text, level, line_num))
headers
() -> [LinkInfo]:
links = []
line_num, line (.lines, ):
link_pattern =
re.finditer(link_pattern, line):
text = .group()
url = .group()
column = .start() +
link_type = ._determine_link_type(url)
links.append(LinkInfo(
text=text,
url=url,
line_number=line_num,
column=column,
link_type=link_type
))
image_pattern =
re.finditer(image_pattern, line):
alt = .group()
url = .group()
column = .start() +
links.append(LinkInfo(
text=alt,
url=url,
line_number=line_num,
column=column,
link_type=
))
links
() -> :
url.startswith():
url.startswith() url.startswith():
url.startswith():
:
() -> :
toc_indicators = [, , , , ]
(indicator .content indicator toc_indicators)
() -> :
todo_patterns = [, , , , ]
(re.search(pattern, .content, re.IGNORECASE) pattern todo_patterns)
() -> :
(re.findall(, .content))
() -> :
(re.findall(, .content))
() -> :
table_lines = [line line .lines line]
([line line table_lines line.strip().startswith()])
() -> :
([line line .lines re.(, line) re.(, line)])
() -> :
fences = re.findall(, .content, re.MULTILINE)
(fences) % ==
() -> :
() -> :
() -> :
header_lines = [i i, line (.lines) re.(, line)]
proper_spacing =
line_num header_lines:
line_num > .lines[line_num - ].strip() == :
proper_spacing +=
(header_lines) > proper_spacing / (header_lines) >
() -> :
words = re.findall(, .content.lower())
([word word words (word) > ])
() -> :
words = re.findall(, .content.lower())
(words) < :
word_counts = Counter(words)
most_common = word_counts.most_common()
most_common[][] / (words) >
() -> :
total_score =
total_weight =
metric .metrics:
category =
metric.name:
category =
metric.name:
category =
metric.name:
category =
metric.name:
category =
metric.name:
category =
category category .weights:
weight = .weights[category]
percentage = metric.score / metric.max_score
total_score += percentage * weight *
total_weight += weight
total_score total_weight >
() -> :
score >= :
score >= :
score >= :
score >= :
:
() -> []:
recommendations = []
metric .metrics:
metric.score < metric.max_score * :
metric.name:
recommendations.append()
metric.name:
recommendations.append()
metric.name:
recommendations.append()
metric.name:
recommendations.append()
metric.name:
recommendations.append()
recommendations
__name__ == :
argparse
parser = argparse.ArgumentParser(description=)
parser.add_argument(, =)
parser.add_argument(, =)
parser.add_argument(, choices=[, ], default=, =)
args = parser.parse_args()
scorer = DocumentQualityScorer(args.file)
:
result = scorer.score_document()
args. == :
args.output:
(args.output, , encoding=) f:
json.dump(result, f, indent=, ensure_ascii=)
()
:
(json.dumps(result, indent=, ensure_ascii=))
:
result:
()
exit()
( * )
()
( * )
()
()
()
()
()
metric result[]:
()
()
()
metric[]:
()
issue metric[]:
()
()
result[]:
()
i, rec (result[], ):
()
()
grade_descriptions = {
: ,
: ,
: ,
: ,
:
}
()
()
Exception e:
()
exit()