skill_satire_news_scraper.md
4.61 KB
Satire News Scraper Skill
Metadata
skill_id: satire_news_scraper
name: Satire News Scraper
version: "1.0.0"
type: content_extraction
domain: satire_news / comedy_materials
target_score: 75
current_score: 80
status: enabled
Description
抓取讽刺新闻网站 (The Onion, ClickHole, Babylon Bee),识别荒诞内容,提取短剧创作素材。
Capabilities
- RSS Feed 解析与抓取
- 荒诞度评分算法
- 讽刺角度识别
- 结构化 JSON 输出
Input Schema
{
"type": "object",
"required": ["source"],
"properties": {
"source": {
"type": "string",
"enum": ["theonion", "clickhole", "babylonbee", "all"]
},
"limit": {
"type": "integer",
"default": 20,
"maximum": 50
},
"min_absurdity_score": {
"type": "number",
"default": 6.0,
"minimum": 0,
"maximum": 10
}
}
}
Output Schema
{
"type": "object",
"properties": {
"results": {
"type": "array",
"items": {
"type": "object",
"properties": {
"headline": "string",
"absurd_premise": "string",
"satire_angle": "string",
"dramatic_twist": "string",
"adaptation_notes": {
"genre": "string",
"tone": "string",
"visual_elements": ["string"],
"dialogue_style": "string"
},
"metadata": {
"source": "string",
"url": "string",
"published_at": "datetime",
"absurdity_score": "number"
}
}
}
}
}
}
Execution Logic
Step 1: RSS Fetch
def fetch_rss(source):
urls = {
"theonion": "https://www.theonion.com/rss",
"clickhole": "https://www.clickhole.com/rss",
"babylonbee": "https://babylonbee.com/feed"
}
response = requests.get(urls[source], headers={"User-Agent": "SatireScraper/1.0"})
return parse_xml(response.content)
Step 2: Content Parsing
def parse_item(item):
return {
"headline": item.find("title").text,
"summary": clean_html(item.find("description").text),
"url": item.find("link").text,
"published_at": parse_date(item.find("pubDate").text),
"source": detect_source(item)
}
Step 3: Absurdity Scoring
def calculate_absurdity_score(headline):
score = 0
# Hyperbole keywords (30%)
hyperbole = ["million", "billion", "infinity", "100%", "everyone", "nobody", "always", "never"]
matches = sum(1 for h in hyperbole if h.lower() in headline.lower())
score += min(matches * 3, 30)
# Absurd subjects (25%)
absurd_patterns = [
r"(Local|Area) (Man|Woman)",
r"(Dog|Cat|Baby|Goose)",
r"(Congress|Senate|President)"
]
for pattern in absurd_patterns:
if re.search(pattern, headline, re.I):
score += 8
# Action absurdity (25%)
absurd_actions = ["declares", "announces", "elects", "appoints", "files", "passes"]
for action in absurd_actions:
if action.lower() in headline.lower():
score += 5
# Structural irony (20%)
irony_indicators = ["actually", "shocked", "surprised", "unexpectedly"]
for indicator in irony_indicators:
if indicator.lower() in headline.lower():
score += 5
return min(score, 100) / 10
Step 4: Content Extraction
def extract_satire_elements(headline, summary):
return {
"absurd_premise": extract_premise(headline),
"satire_angle": detect_satire_target(headline, summary),
"dramatic_twist": identify_twist(headline),
"adaptation_notes": generate_adaptation_notes(headline)
}
Step 5: JSON Output
def generate_output(parsed, score, elements):
return {
"headline": parsed["headline"],
"absurd_premise": elements["absurd_premise"],
"satire_angle": elements["satire_angle"],
"dramatic_twist": elements["dramatic_twist"],
"adaptation_notes": elements["adaptation_notes"],
"metadata": {
"source": parsed["source"],
"url": parsed["url"],
"published_at": parsed["published_at"].isoformat(),
"absurdity_score": round(score, 1)
}
}
Error Handling
- RSS 不可达 → 返回错误,建议检查网络
- 解析失败 → 跳过该条目,继续处理
- 编码问题 → 强制 UTF-8,失败则标记
Example Usage
curl -X POST http://localhost:3000/api/agent/task/dispatch \
-d '{
"to": "satire-news-scraper",
"task": "抓取 The Onion 最新内容,筛选 absurdity_score > 7.0"
}'