Analyze Article

Analyze Article

Analyze news articles using LLM to extract insights and assess importance.

When to Use

Use this skill when you need to:

• Generate concise summaries of news articles
• Categorize articles by topic (research, business, product, security, policy)
• Extract key entities (companies, people, technologies, models)
• Assess article importance on a 1-10 scale
• Detect breaking news that requires immediate notification

Instructions

Step 1: Retrieve Article from Memory

Use memory/get to retrieve the article to analyze by its ID.

The article should contain:

• title: Article title
• url: Source URL
• source: Publication name
• summary or content: Article text

Step 2: Analyze with LLM

Use the use_llm tool to analyze the article with this prompt structure:

Analysis prompt:

Analyze the following news article and provide:

1. **Summary**: A concise 2-3 sentence summary highlighting the key points.
2. **Category**: One of: research, business, product, security, policy, general
3. **Entities**: List of key entities mentioned (companies, people, technologies, models)
4. **Importance Score**: 1-10 rating where:
   - 1-3: Minor news, incremental updates
   - 4-6: Notable news, meaningful developments
   - 7-8: Important news, significant impact
   - 9-10: Major news, industry-changing announcements
5. **Is Breaking**: True if this is major breaking news
6. **Breaking Reason**: If breaking, explain why

Article:
Title: {title}
Source: {source}
Content: {content}

Respond in JSON format.

Importance scoring factors:

• Source credibility and significance
• Novelty of the information
• Potential industry impact
• Whether from an official company announcement
• Security implications

Step 3: Store Analysis in Memory

Use memory/add to store the analysis results:

• type: "analysis"
• namespace: "news/analyses"
• data: {summary, category, entities, importance_score, is_breaking, breaking_reason}
• metadata: {source_article_id, analyzed_at}

Step 4: Link Analysis to Source

Use memory/link to create a relationship:

• source_id: analysis ID
• target_id: original article ID
• relation_type: "ANALYZED_FROM"

Step 5: Return Results

Return the analysis including:

• Analysis ID for reference
• AI-generated summary
• Category classification
• Extracted entities
• Importance score
• Breaking news flag

Tool Usage Guidance

use_llm tool

• Use for structured content analysis
• Request JSON output format
• Use temperature 0.0 for consistency

memory/get

• Retrieve article by ID
• Returns full article data

memory/add

• Store analysis as type "analysis"
• Include source article ID in metadata

memory/link

• Create ANALYZED_FROM relationship
• Links analysis to source article

Importance Scoring Guidelines

Score 1-3: Minor News

• Incremental product updates
• Minor bug fixes or patches
• Routine announcements

Score 4-6: Notable News

• New features or capabilities
• Meaningful partnerships
• Research paper publications

Score 7-8: Important News

• Major product launches
• Significant research breakthroughs
• Important policy changes

Score 9-10: Major News

• Industry-changing announcements
• Major security vulnerabilities
• Breakthrough research results

Breaking News Criteria

Mark as breaking if ANY of these apply:

• Major model release from leading AI companies (OpenAI, Anthropic, Google, Meta)
• Critical security vulnerability affecting widely-used AI systems
• Regulatory action with immediate industry impact
• Breakthrough research that changes fundamental understanding

Analysis Data Schema

{
  "id": "analysis-abc123",
  "summary": "OpenAI has released GPT-5 with significant improvements...",
  "category": "product",
  "entities": ["OpenAI", "GPT-5", "Sam Altman"],
  "importance_score": 9,
  "is_breaking": true,
  "breaking_reason": "Major model release from leading AI company",
  "source_article_id": "article-xyz789",
  "analyzed_at": "2026-01-31T12:00:00Z"
}

Error Handling

• If article retrieval fails, log error and skip
• If LLM returns malformed JSON, retry with clearer prompt
• If memory operations fail, log but return analysis results

Success Criteria

• All articles have valid summaries and categories
• Importance scores are calibrated and consistent
• Entities are accurately extracted
• Analysis is linked to source article