agentleFS
vanzan01vanzan01/claude-agent-sdk-starter/.claude/skills/news-tools/SKILL.md

news-tools

Skill9 starsChanged 10 months agoMITView on GitHub

Utilities for fetching and structuring AI news items for multi-agent pipelines. Use when gathering AI headlines, analyzing news, or crafting tweets in the ai-news-tweet app.

The file

We split this skill into the rules it gives an agent

12 rules. About 0 are common, 9 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈9

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    When you (Claude) need AI news in this app, you should always use WebSearch to get real news from TODAY ONLY.

    line 16
  2. Distinctive:

    ONLY include news published TODAY: Do NOT include articles from yesterday or earlier.

    line 20
  3. Distinctive:

    If the search returns older articles, REJECT them

    line 22
  4. Very specific:

    Only articles with published_at matching today's date are valid

    line 23
  5. Distinctive:

    If no news from today is found, say so - do NOT use older articles

    line 24
  6. Very specific:

    When the user asks for AI news or when you are running inside the ai-news-tweet pipeline:

    line 28
  7. Distinctive:

    Search for: "AI news today [current date]" or "AI news [YYYY-MM-DD]"

    line 31
  8. Distinctive:

    Do NOT use your world knowledge or make up headlines

    line 34
  9. Distinctive:

    Use real URLs from the search

    line 56
  10. Very specific:

    If your system prompt specifies output markers (like <<<researcher>>>...<<<end-researcher>>>), wrap your JSON output in those markers

    line 59
  11. Distinctive:

    Do NOT fabricate news as a fallback

    line 64
  12. Distinctive:

    Do NOT fall back to older articles

    line 65

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Overview

news-tools is the AI news skill for the ai-news-tweet app. It provides the capability to fetch real, current AI news using WebSearch.

  • Primary user: the researcher subagent in ai-news-tweet.
  • Primary output: structured JSON with AI news items that downstream agents can analyse and rephrase.

When you (Claude) need AI news in this app, you should always use WebSearch to get real news from TODAY ONLY.

CRITICAL: Today's Date Only

ONLY include news published TODAY. Do NOT include articles from yesterday or earlier.

  • If the search returns older articles, REJECT them
  • Only articles with published_at matching today's date are valid
  • If no news from today is found, say so - do NOT use older articles

How Claude Should Use This Skill

When the user asks for AI news or when you are running inside the ai-news-tweet pipeline:

  1. Use WebSearch to get TODAY's news ONLY.
    • Search for: "AI news today [current date]" or "AI news [YYYY-MM-DD]"
    • Get actual headlines from real sources
    • REJECT any articles not published today
    • Do NOT use your world knowledge or make up headlines
  2. Structure the results as JSON using this EXACT format:

    {
      "topic": "ai",
      "items": [
        {
          "title": "Google launches Gemini 3 and Ironwood TPU for AI infrastructure",
          "summary": "Google unveiled Gemini 3 (requiring less prompting) and Ironwood, 7th gen TPU for scaling massive models",
          "source": "CNBC",
          "url": "https://www.cnbc.com/2025/11/27/example-article.html",
          "published_at": "2025-11-27"
        }
      ]
    }
  3. Return 3-5 news items from TODAY.
    • Focus on the most significant/interesting AI developments
    • published_at MUST be today's date in YYYY-MM-DD format
    • Use real URLs from the search
  4. IMPORTANT: Output Markers
    • If your system prompt specifies output markers (like <<<researcher>>>...<<<end-researcher>>>), wrap your JSON output in those markers
    • The markers go OUTSIDE your JSON content
  5. If WebSearch fails or no news from today:
    • Report the error clearly
    • Do NOT fabricate news as a fallback
    • Do NOT fall back to older articles

Tools

WebSearch

  • Purpose: Fetch real AI news from TODAY
  • Usage: Use the WebSearch tool with queries like:
    • "AI news today December 7 2025"
    • "artificial intelligence news 2025-12-07"
    • "latest AI developments today"
  • Output: Real news articles with titles, summaries, sources, and URLs
  • CRITICAL: Only include results published TODAY

Pipeline Architecture

The ai-news-tweet app uses a 3-stage agent pipeline, each with its own specialized skill:

StageAgentSkillPurpose
1researchernews-tools (this skill)Fetch real AI news via WebSearch
2analysisanalysis-helperPick most impactful story
3writertweet-writerCraft engaging tweet

Each stage passes its output to the next via structured markers.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.