digest

The digest skill is a specialized agent capability designed to transform overwhelming information streams into concise, high-signal summaries. It is built for developers, researchers, and professionals who need to stay updated on tech news, AI developments, or academic breakthroughs without wasting time on clickbait or low-value content. By enforcing a strict policy of using curated, trusted sources, this skill ensures that the information delivered is both accurate and relevant to the user's specific domain of interest.

• Automatically fetches the latest news, daily or weekly roundups, and research paper summaries based on user-defined topics.

• Implements a strict quality filter that prioritizes technical outlets like Hacker News, Ars Technica, arXiv, Semantic Scholar, Papers with Code, and OpenReview.

• Produces a standardized, scannable markdown format featuring headlines, key takeaways, source attribution with direct links, and optional deep-dive summaries.

• Supports time-aware retrieval, allowing users to specify windows such as 'this week,' 'last 72 hours,' '2024,' or specific publication dates to ensure temporal relevance.

• Simplifies complex topics like 'what is new in AI' or 'WebAssembly trends' by synthesizing data from multiple credible sources into one coherent report.

• Highly customizable through user preferences, allowing for the exclusion of specific unwanted topics (like crypto) or the addition of custom trusted outlets.

• When invoking the skill, clearly define the topic and the desired time frame; if no time is specified, the agent defaults to the most recent updates.

• The output is structured to be brief—prefer one-line summaries unless a deep dive is requested, keeping the information dense and highly readable.

• Always verifies sources against the internal curated list to maintain integrity and reduce hallucinations or low-signal input.

• Input expected includes natural language requests like 'latest tech news,' 'weekly AI research roundup,' or 'what is happening with Rust this month.'

• Practical constraints: The tool is optimized for professional-grade news and academic research; it is not designed for real-time social media sentiment analysis or generic, non-curated news aggregation.