How to Automate Daily News Briefings with OpenClaw: Skill vs Prompt

Using Skill to Create Briefings

In OpenClaw, a Skill is a reusable package that encapsulates a specific workflow, similar to a plugin but focused on a single task. Skills are easy to configure and share, making them suitable for straightforward news‑briefing automation.

Configuration Steps

Create the news Skill

Trigger condition : when to run (e.g., every day at 08:00).

Data source : list of RSS feeds.

Processing logic : filter, categorize, and summarize articles.

Output format : template for the briefing.

Configure RSS sources

# news skill configuration example
sources:
- name: "Tech News"
  url: "https://www.36kr.com/feed"
- name: "Developer Updates"
  url: "https://sspai.com/feed"
- name: "AI Frontier"
  url: "https://www.artificialintelligence-news.com/feed/"

Set the execution schedule

schedule:
  cron: "0 8 * * *"   # run daily at 08:00
  timezone: "Asia/Shanghai"

Result

When the configuration is saved, OpenClaw executes the Skill automatically and performs the following actions:

Fetches the latest items from the configured RSS feeds.

Uses an AI model to filter irrelevant content and generate concise summaries.

Builds a structured briefing according to the defined template.

Pushes the briefing to the selected channel (e.g., Feishu, QQ).

Pros and Cons

Pros

Simple configuration, ready‑to‑use.

Reusable across projects and easy to share.

Stable execution with low error rate.

Cons

Limited flexibility for complex logic.

Any change requires redeployment.

Hard to express sophisticated filtering rules.

Using Prompt to Create Timed Briefings

The Prompt approach relies on a custom prompt template that tells OpenClaw how to process news. This method is more adaptable for personalized or rapidly changing requirements.

Configuration Steps

Create the Prompt template

You are my news assistant, responsible for daily tech news.

Task:
1. Pull today’s articles from the following RSS sources:
   - 36Kr: https://www.36kr.com/feed
   - 少数派: https://sspai.com/feed

2. Filtering criteria:
   - Prioritize AI, programming tools, productivity.
   - Exclude pure ads and soft‑sell articles.
   - No more than 5 items per category.

3. Output format:
## 📰 Daily Tech Briefing {{date}}

### 🤖 AI Frontier
- [Title](Link "Title") - One‑sentence summary

### 💻 Development Tools
- [Title](Link "Title") - One‑sentence summary

Please format today’s news accordingly.

Configure the scheduled task

prompts:
- id: "daily-news-briefing"
  name: "Daily Tech Briefing"
  trigger:
    type: "schedule"
    cron: "0 8 * * *"
    timezone: "Asia/Shanghai"
  content: |
    (Prompt template from step 1)
  output:
    channel: "feishu"
    format: "markdown"

Test and refine

# Test Prompt
openclaw prompt test daily-news-briefing

Iterate on the prompt until the generated briefing matches expectations.

Result

After a successful test, the Prompt runs on schedule and produces a markdown briefing such as:

## 📰 Daily Tech Briefing 2025-02-19

### 🤖 AI Frontier
- Anthropic releases Claude 4 – inference speed up 300%
- OpenAI unveils GPT‑5 preview – multimodal input support

### 💻 Development Tools
- VS Code new features – built‑in AI code review
- Rust 2.0 released – major performance boost

### 📱 Productivity
- Notion AI update – Chinese document generation

Pros and Cons

Pros

Highly customizable; quick to iterate.

Can express complex filtering logic that Skill cannot.

Changes take effect immediately without redeployment.

Cons

Requires proficiency in prompt engineering.

Needs ongoing debugging and tuning.

Stability depends on the quality of the prompt.

Choosing Between Skill and Prompt

Scenario: Quick start, basic needs → Recommend Skill because it is simple and works out‑of‑the‑box.

Scenario: Highly personalized briefing → Recommend Prompt for flexible customization.

Scenario: Team‑wide shared configuration → Recommend Skill for easy distribution and reuse.

Scenario: Frequent content adjustments → Recommend Prompt as changes are immediate.

Scenario: Complex filtering logic → Recommend Prompt because AI interpretation handles nuance better.