AI Daily

Problem

Large models have training cutoffs, so an agent cannot answer “what is happening in AI today” or “which model releases happened in the last 24 hours” from memory. AI HOT provides daily AI updates, but natural-language queries are ambiguous: broad questions, explicit daily reports, full feeds, category filters, and keyword searches need different routing. This skill constrains the agent’s decisions to avoid defaulting to full results, misusing the daily page, or dumping raw API details to users.

How it works

The skill first interprets intent: broad questions default to selected items, optionally with a recent time window; only explicit “daily report” requests use the edited daily page; only explicit “all/full/everything” requests return unselected entries. It supports the five categories ai-models, ai-products, industry, paper, and tip, plus keyword search and result limits. Output is grouped by section and preserves title, source, human-readable time, summary, and link, making it easier for engineers to read or post-process.

Boundaries

The recent query window covers the last 7 days, and result size is capped at 100. Older content should be read from dated archives. The skill is suitable for current AI industry facts, not for paper close reading, competitive analysis, or long-term trend research. For non-technical users, output should avoid exposing endpoints, parameters, rate limits, or other internal details.