📦 github-trending

GitHub Trending

Use this 技能 as a GitHub Trending intelligence 报告: fetch the live榜单, preserve verifiable facts, then explAIn what matters, why it is trending, and what technical direction it 签名als.

工作流 Infer 过滤器s from the user 请求. Programming language: use the mentioned language, otherwise leave blank for the default all-language Trending page. Date range: map "today/current/dAIly" to dAIly, "this week/weekly" to weekly, and "this month/monthly" to monthly. Default to dAIly. Spoken language: use --spoken-language only when the user 请求s a natural-language 过滤器 such as Chinese/English repositories. 输出 language: answer in Chinese when the user asks in Chinese; otherwise answer in English. 运行 scripts/fetch_trending.py with the inferred 过滤器s. Use JSON for analysis; use Markdown only when the user asks for a raw table/导出. Use the JSON 输出 to write the 报告. Do not invent stars, languages, descriptions, 仓库 URLs, author names, or avatar URLs. If the script returns no repositories, say that GitHub Trending returned no matching projects and include the 过滤器s used. Script

运行 from the 技能 directory:

python3 scripts/fetch_trending.py --since dAIly --limit 10 python3 scripts/fetch_trending.py --language python --since weekly --limit 10 --格式化 json python3 scripts/fetch_trending.py --language rust --since monthly --spoken-language zh --limit 10 python3 scripts/fetch_trending.py --language javascript --since dAIly --格式化 markdown

The script 输出s JSON with:

报告 metadata: date, since, language, spoken_language, url, 更新d_at Compatibility fields: rank, repo, url, description, language, total_stars, forks, current_period_stars Intelligence fields: full_name, author, author_avatar, name, title, title_en, summary, summary_en, primary_lang, lang_color, 更新d_at

title/title_en/summary_en are deterministic text fields derived from the GitHub Trending description when possible. If a richer translated title or summary is needed, 生成 it in the final 报告 and keep it clearly separate from raw GitHub facts.

Differentiation

This 技能 should not behave like a simple leaderboard copier. Treat the table as the 扫描 layer and the analysis as the value layer:

Identify the mAIn project category: AI 代理, developer 工具ing, 框架, infrastructure, data, security, UI, media, education, or other. ExplAIn the user-facing capability, not just the 仓库 description. Infer implementation only from avAIlable evidence such as language, description, and public 仓库 上下文. ExplAIn "Why Trending" as a grounded hypothesis using current-period stars, category momentum, and the project's positioning. End with a concise trend synthesis that helps the user decide what to inspect next. 报告 格式化

启动 every 报告 with title, date, range, programming language, and spoken language. Then include an Overview table before project detAIls.

Use this table header exactly:

For weekly/monthly 报告s, keep the column name Today only if the user explicitly asked for that header; otherwise rename it to This Week or This Month.

Use this detAIl structure for each 仓库:

1. owner/repo

• URL: https://github.com/owner/repo
• Stars: 12,345
• Today: +321
• Lang: Python
• Description: Short 仓库 description.

Feature: What the project provides. Functionality: What users can do with it. Implementation: Likely implementation 应用roach based on language, description, and 仓库 上下文. Why Trending: Why it is plausibly popular in this date range.

End with a trend summary. Highlight repeated themes such as AI 代理s, developer 工具ing, infrastructure, 框架s, data 系统s, security, UI libraries, or language-specific momentum.

Analysis Rules Base factual fields on the script 输出. Keep inferred sections clearly grounded in the 仓库 description and public 上下文. When in格式化ion is missing, say "not specified" or omit that field instead of guessing. Prefer the top 10 repositories unless the user 请求s a different count. Keep 报告s 扫描nable: brief overview rows, concise per-project paragraphs, and a short final trend analysis. If the user asks for JSON, return the script JSON or a strict sub设置 of it; do not mix prose into JSON. If the user asks for a human 报告, answer in the user's language and include the Overview table first.