news-monitor

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

News Monitor

新闻监控

This skill finds recent news and developments on a topic and delivers a chronological briefing. The focus is on what's new — not a general overview.

此技能可查找某一主题的近期新闻与动态，并按时间顺序提供简报。核心聚焦于最新动态——而非通用概述。

Design principle: search results first, fetch sparingly

设计原则：优先搜索结果，谨慎获取内容

The news search tool already returns titles, snippets, dates, and sources for each result. That's enough to build most of the briefing without fetching any pages. Only fetch a page when a snippet is too vague to understand what actually happened and you need the full article to write a useful summary. Most of the time, 0–2 fetches is plenty. Never fetch more than 3 pages.

新闻搜索工具已返回每条结果的标题、摘要、日期和来源。无需获取页面内容即可构建大部分简报。仅当摘要过于模糊，无法了解实际发生的情况，且需要完整文章来撰写有用的摘要时，才获取页面内容。大多数情况下，0-2次获取已足够。绝对不要获取超过3个页面。

Tools available

可用工具

News search — the primary tool. Returns results sorted by date with timestamps and source outlets.

If an MCP news search tool is available (e.g.,

mcp__web_forager__duckduckgo_news_search

mcp__duckduckgo__duckduckgo_news_search

), prefer it:

mcp__web_forager__duckduckgo_news_search(query="your query", max_results=10)

Each result includes

title

url

snippet

date

, and

source

Without MCP, run the packaged Web Forager CLI with uvx:

bash

uvx --python '>=3.10,<3.14' web-forager news "your query" --max-results 10 --output-format json

uvx

cannot run the packaged CLI, use a direct

ddgs

fallback through uv without touching the current project environment:

bash

uv run --no-project --python '>=3.10,<3.14' --with 'ddgs>=9.5.2' python - <<'PY'
from ddgs import DDGS
results = DDGS().news(query="your query", max_results=10)
for r in results:
    print(r["date"], r["title"], r["url"], r["source"])
PY

General search — use an MCP search tool,

web-forager search

via uvx, or

DDGS().text()

via

uv run --no-project

only if news search returns too few results.

Fetch — use an MCP fetch tool or

curl -s "https://r.jina.ai/URL"

only when a snippet is too vague to summarize the event. Cap fetches with

max_length=3000

to avoid pulling giant pages.

新闻搜索——主要工具。返回按日期排序的结果，包含时间戳和来源媒体。

如果有MCP新闻搜索工具可用（例如

mcp__web_forager__duckduckgo_news_search

或

mcp__duckduckgo__duckduckgo_news_search

），优先使用：

mcp__web_forager__duckduckgo_news_search(query="your query", max_results=10)

每条结果包含

title

、

url

、

snippet

、

date

和

source

。

若无MCP工具，使用uvx运行打包的Web Forager CLI：

bash

uvx --python '>=3.10,<3.14' web-forager news "your query" --max-results 10 --output-format json

若无法通过uvx运行打包的CLI，则通过uv使用直接的

ddgs

回退方案，无需触碰当前项目环境：

bash

uv run --no-project --python '>=3.10,<3.14' --with 'ddgs>=9.5.2' python - <<'PY'
from ddgs import DDGS
results = DDGS().news(query="your query", max_results=10)
for r in results:
    print(r["date"], r["title"], r["url"], r["source"])
PY

通用搜索——仅当新闻搜索返回结果过少时，才使用MCP搜索工具、通过uvx运行

web-forager search

，或通过

uv run --no-project

使用

DDGS().text()

。

内容获取——仅当摘要过于模糊无法总结事件时，才使用MCP获取工具或

curl -s "https://r.jina.ai/URL"

。设置

max_length=3000

来限制获取内容，避免加载大页面。

News monitoring workflow

新闻监控工作流程

Step 1 — Determine scope

步骤1 — 确定范围

Before searching, clarify:

Topic: what exactly are we monitoring?
Time frame: did the user specify "this week", "last month", "since X"? If not, default to the last 2–4 weeks.
Angle: are they interested in everything, or a specific aspect (e.g., "funding news about X", "regulatory updates on Y")?

搜索前，明确以下内容：

主题：我们要监控的具体内容是什么？
时间范围：用户是否指定了“本周”、“上月”、“自X以来”？若未指定，默认过去2-4周。
角度：用户是否对所有内容感兴趣，还是特定方面（例如“X的融资新闻”、“Y的监管更新”）？

Step 2 — Search

步骤2 — 搜索

Run 1–2 news searches with different angles to get good coverage:

```
"[topic]"
```
as the base query
A second query with a more specific angle if relevant (e.g., "[topic] announcement", "[topic] policy", "[topic] release")

Include the current year in queries to bias toward recent results.

运行1-2次不同角度的新闻搜索以获取全面覆盖：

基础查询为
```
"[topic]"
```
若相关，可使用更具体角度的二次查询（例如“[topic] 公告”、“[topic] 政策”、“[topic] 发布”）

在查询中包含当前年份，以偏向近期结果。

Step 3 — Deduplicate and filter

步骤3 — 去重与筛选

From the combined news results:

Group articles about the same event (multiple outlets covering the same story)
Pick the best source for each event (prefer the one with the most informative snippet)
Discard anything outside the relevant time frame
Discard duplicates or near-duplicates
Aim for 3–7 distinct events

从合并的新闻结果中：

将报道同一事件的文章分组（多个媒体报道同一新闻）
为每个事件选择最佳来源（优先选择摘要信息最丰富的来源）
剔除超出相关时间范围的内容
剔除重复或近乎重复的内容
目标保留3-7个不同事件

Step 4 — Selective fetch (only if needed)

步骤4 — 选择性获取（仅在需要时）

Look at your filtered results. For each event, ask: "Can I write a useful 2–4 sentence summary from the title + snippet alone?" If yes, don't fetch. If the snippet is cryptic or you need key details (numbers, names, outcomes), fetch that one page with

max_length=3000

查看筛选后的结果。对于每个事件，问自己：“仅通过标题+摘要，我能否撰写一段有用的2-4句摘要？”如果可以，无需获取。如果摘要晦涩难懂，或需要关键细节（数字、名称、结果），则使用

max_length=3000

获取该页面。

Step 5 — Deliver the briefing

步骤5 — 交付简报

Organize chronologically (most recent first) and present as a news briefing.

按时间顺序（最新优先）整理，并以新闻简报的形式呈现。

Output format

输出格式

undefined

undefined

[Topic] — News Briefing

[主题] — 新闻简报

Period: [time frame covered] Last updated: [today's date]

周期：[涵盖的时间范围] 最新更新：[今日日期]

Headlines

头条新闻

[One-line summary of event 1] — [date]
[One-line summary of event 2] — [date]
[One-line summary of event 3] — [date]

[事件1的一句话摘要] — [日期]
[事件2的一句话摘要] — [日期]
[事件3的一句话摘要] — [日期]

Details

详细内容

[Event 1 title] — [date]

[事件1标题] — [日期]

[2–4 sentences: what happened, why it matters, what's next] Source: Title

[2-4句话：发生了什么、重要性、后续发展] 来源：标题

[Event 2 title] — [date]

[事件2标题] — [日期]

[2–4 sentences] Source: Title

[2-4句话] 来源：标题

[Event 3 title] — [date]

[事件3标题] — [日期]

[2–4 sentences] Source: Title

[2-4句话] 来源：标题

What to watch

值得关注的内容

[1–2 sentences about upcoming events, expected announcements, or unresolved threads that the user might want to follow up on]


Keep it tight. The user wants to catch up quickly, not read essays. If there's genuinely
no recent news, say so — "No significant developments found in [time frame]" is a valid
and useful answer.

[1-2句话：用户可能想要跟进的即将发生的事件、预期公告或未解决的事项]


保持内容简洁。用户希望快速了解情况，而非阅读长篇大论。若确实没有近期新闻，直接说明——“在[时间范围]内未发现重大动态”是有效且有用的回复。