ai-newsletter（艾时事通讯）

AI Newsletter Daily

Generate a concise daily AI newsletter from fresh web sources.

Use this skill only for current AI/ML news, releases, research, funding, product launches, model updates, regulation, benchmarks, or practitioner-relevant developments.

Do not use for evergreen explainers, non-AI topics, or long-form research that is not intended to become a curated newsletter.

Inputs

Defaults:

• target_news_count = 20
• search_query = "latest AI news today"
• search_time_window_days = 2
• max_search_results = 60
• min_articles_required = 10
• include_domains = []
• exclude_domains = ["youtube.com", "reddit.com", "facebook.com", "x.com", "twitter.com"]
• summary_model = "host-default"
• max_scrape_retries = 2

Bounds:

• target_news_count: 1..50
• search_time_window_days: 1..14
• max_search_results: 20..120
• min_articles_required: 1..50
• max_scrape_retries: 0..5

If min_articles_required > target_news_count, set it to target_news_count.

Batch policy

• Search up to max_search_results candidates.
• Keep the top target_news_count * 2 candidates for fetch attempts.
• Return only the top target_news_count verified items.
• Do not summarize every search result.

Required outputs

Return:

1. newsletter_items as a list of objects.
2. markdown_newsletter as a string.
3. json_newsletter as an object.

Each item must include:

• title
• url
• domain
• published_at
• summary
• relevance_score
• source_query

Use "unknown" for missing published_at.

Workflow

1. Resolve inputs.

- Apply defaults and bounds. - Initialize warnings = [], seen_canonical_urls = set(), processed_urls = set().

1. Search.

- Run web_search with search_query. - If no usable results, retry once with: - "{search_query} generative AI LLM model open source enterprise" - If still no usable results, fail clearly.

1. Normalize and filter.

- Keep only results with non-empty title and URL. - Canonicalize URLs: lowercase host, remove tracking parameters, normalize safe trailing slashes. - Drop duplicates by canonical URL. - Apply include_domains and exclude_domains. - Prefer results likely within search_time_window_days. - Keep unknown dates, but score them lower.

1. Rank.

- Score each candidate from 0 to 100: - AI-topic relevance: 0..50 - Freshness: 0..30 - Title/snippet clarity: 0..20 - Sort by: - relevance_score desc - published_at desc, unknown last - url asc - Keep the top target_news_count * 2 candidates.

1. Verify and summarize.

- Process candidates in ranked order until target_news_count verified items are collected. - Skip candidates whose canonical URL is already in processed_urls. - Attempt web_fetch up to max_scrape_retries + 1 times. - If fetch fails, add a warning with the URL and reason, then continue. - Cross-check search result vs fetched page using: - title similarity, - domain consistency, - topic alignment, - published date when available. - If the page appears materially inconsistent, skip it and warn. - Summarize in one plain-text paragraph, max about 80 words. - Focus on why it matters to AI practitioners. - If summary generation fails, warn and continue. - Append the enriched item.

1. Minimum quality gate.

- If collected items are fewer than min_articles_required, run one fallback search with: - "AI news today machine learning model release funding research" - Process only new candidates not already seen or processed. - Repeat filtering, ranking, verification, and summarization.

1. Final integrity check.

- Ensure every final item has non-empty title, url, domain, summary, source_query, and numeric relevance_score. - Ensure each URL appears once. - Ensure markdown_newsletter and json_newsletter match in item count. - Remove and warn on any invalid item.

1. Finalize.

- Sort by relevance_score desc, then published_at desc. - Truncate to target_news_count. - Render markdown_newsletter. - Assemble json_newsletter. - Return all outputs.

Verification rules

Accept an item only if it passes these checks:

• URL integrity:

- canonical URL is valid, - duplicates removed, - malformed URLs rejected.

• Source consistency:

- search title and fetched title broadly match, - snippet and page content describe the same story, - off-topic pages rejected.

• Metadata sanity:

- valid published date preferred, - unknown date allowed only if the rest is strong, - malformed or impossible dates rejected.

• Content integrity:

- fetched content must be substantively about the same AI news item, - truncated or malformed pages rejected.

• Warning log:

- record every failed URL and reason, - record whether fallback search was used.

Markdown format

markdown_newsletter must use:

• H1 title with date.
• One H2 section per article.
• One short summary paragraph per article.
• One source link per article.

Example:

AI Newsletter Daily — 2026-04-28

1. Article title

Summary paragraph.

Source: link

Warnings

Only include this section when needed.

Failure policy

Hard fail only when:

• Both initial and fallback searches return no usable URLs.
• Required tools are unavailable.

Soft fail and continue when:

• A single fetch fails.
• A single summary fails.
• published_at is missing.
• A candidate fails cross-check verification.

Partial success is acceptable when the result count is between min_articles_required and target_news_count.

Always include actionable warnings with URL, short reason, and whether fallback search was used.

Safety rules

• Use only sanctioned tools.
• Do not request API keys from the user.
• Do not expose secrets.
• Do not include copyrighted full article text.
• Keep summaries neutral, concise, and factual.
• Preserve deterministic behavior wherever tool outputs allow.

Return shape

json_newsletter must contain:

• date
• query
• count
• articles
• warnings