News Skill: Keep Your Agent Current Without Piling Up APIs

Where this fits

Your agent is mid-task—say, drafting a market brief—and the user asks for "the latest on China's tech policy." You could scrape a dozen sites, but that breaks quickly and returns noise. The news skill solves this by giving your agent a curated list of authoritative news URLs across politics, finance, society, world, tech, sports, and entertainment. It uses browser_use to open each page, snapshot the content, and summarize the key points back to the user—all without a paid news API or fragile custom scrapers.

Why agents benefit

• Zero API keys: No news service subscriptions or OAuth flows; just browse public pages like a human would.
• Category-aware routing: The skill maps user intent ("what's happening in sports?") directly to a trusted source URL, so the agent doesn't have to guess or search.
• Structured extraction: After browser_use snapshots the page, the skill's prompt guides the agent to pull headlines and key facts, not just dump raw HTML.
• Authoritative sources: The curated list includes People's Daily, China Economic Net, CGTN, and Science and Technology Daily—reducing hallucination risk from random blogs.
• Lightweight integration: It's a single skill folder with a README and prompt template; no heavy dependencies beyond browser_use.

Practical scenarios

Morning briefing for a business user

The user says, "Give me today's finance and world news in 5 bullets." The agent opens http://www.ce.cn/ and https://www.cgtn.com/, snapshots both, and synthesizes a concise update with links back to the original articles.

Tech trend monitoring for a developer

A developer asks, "What's new in Chinese tech this week?" The skill routes to Science and Technology Daily, extracts headline trends, and highlights one or two stories that might affect their stack—without the agent having to maintain a custom RSS feed.

Real-time context during a research task

While researching a policy topic, the user interjects, "Check the latest from CPC News." The agent switches context, opens https://cpc.people.com.cn/, and pulls the top political headlines to enrich the ongoing analysis.

Add it to your agent workflow

Install the skill via the OpenClaw CLI:

npx skills add hyqibot/token-free-openclaw --skill news

Then, in your agent's system prompt or skill trigger, reference it like this:

When the user asks for news, use the `news` skill. Open the category URL with browser_use, snapshot the page, and summarize the top 3-5 headlines with source links.

The skill's README includes the full category-to-URL table, so your agent can route correctly without extra configuration.

Compare before adopting

Before you commit, weigh the skill against alternatives:

• Quality signals: Check the GitHub repo's issue tracker and star history (currently 121 stars). Are sources updated when sites change structure?
• Maintenance freshness: The README lists a version (1.1) and a copaw emoji, but verify the last commit date—news URLs rot fast.
• Alternatives: Compare with dedicated news APIs (e.g., NewsAPI, GDELT) if you need historical archives or multilingual coverage beyond Chinese sources.
• Workflow fit: This skill is ideal for lightweight, on-demand news pulls. For high-frequency polling, a dedicated API or RSS pipeline may be more reliable.

Why it is worth tracking

This skill stands out because it solves a real pain point—current, authoritative news without a subscription—using a browser-based approach that's flexible and transparent. The project is young but active, and the community's willingness to share curated source lists is a good sign. Track it if you're building agents that need to stay contextually aware of world events, especially those with a China focus. Evaluate it when your agent's news capability becomes a bottleneck or you want to avoid third-party API dependencies. For more context on how this fits the broader skill ecosystem, see OpenAgentSkill's directory entry.