Description

Suitable for creators preparing to write, research, or select content topics. Provide the topic, audience, purpose, region/language, source limits, and deadline; defaults to 30 calendar days including the deadline in Beijing time. Specify other time zones or historical windows as needed. Process: Determine the window and queries, open accessible primary sources, and verify event occurrence dates, publication dates, update dates, and access dates. Merge syndicated coverage of the same event and different versions, distinguish official announcements, discussions, and inferences, and then provide evidence-backed topic angles and verification actions. When no recent credible information is available, leave the results empty rather than using old news to pad counts. Delivery: A Chinese topic briefing, an evidence table with URLs, locations, dates, and publication statuses, and explanations of candidate exclusions, failures, and source coverage. If file capability is available, save as Markdown, CSV, or JSON; otherwise, provide copyable content. You can trace back to the original text along with the entries, and then test functions or user feedback that may affect judgment; suggestions do not imply execution. Four local real query cases: AI models, coding assistants, code video tools, and knowledge management updates. With 2026-11-11 as the deadline and a window from 2026-10-13 to 11-11, actual official announcements and release notes were read to organize 8 update records with clear dates. GitHub was used to fetch official release JSON for exact publication times and Shanghai dates. Obsidian early access and public versions were treated separately, and out-of-window old releases or same-event versions were excluded. This is a self-devised research requirement, not a client project. These four cases limit official update materials and did not acquire social media comments or comparable interaction time series. They are not whole-web popularity rankings, public opinion surveys, or public consensus; multiple pages from the same organization do not count as independent consensus. Candidate software was not installed, upgraded, or performance-tested, and official performance descriptions do not equal personal measurements. Local image tool models cannot be determined based on new model announcements. Four 1920x1080 displays show original typography excerpts of the actual briefings, with complete records saved locally; individual avatars are ImageGen conceptual visuals, not news or popularity evidence. Requires a text model (GPT recommended) and real internet search or webpage reading capabilities. Social media authorization, subtitles, PDF/OCR, and file exports depend on environment capabilities. A vision model is not strictly required, and the skill does not include an upstream last30days engine, paid APIs, or accounts. When platform tools are missing, only provided materials are processed with a note that real-time queries were not performed; stories cannot be made up without sources. YouMind internet/file capabilities and fixed model bindings were verified without consuming points. 199 points is the installation price and does not include model/API or third-party service fees. Cookies, registration, payments, posting, or scheduled crawling are not automatically handled, and exhaustive coverage, traffic, paid conversion, revenue, or moderation are not promised. Likes, comments, views, installs, and orders are not mixed together, paid rates without a denominator are not calculated, and cumulative totals are not treated as 30-day increments. Materials may change, so recheck before formal use.