skillnav（skillnav 搜索）

<!-- skill_version: 2.0.0 -->

name: skillnav description: "Search 3,900+ MCP servers with install commands, get daily AI brief, query papers, search everything, and discover trending tools — in Chinese. Data from skillnav.dev editorial team." version: 2.0.0 argument-hint: "brief | mcp <keyword> | search <keyword> | paper <id|keyword> | trending | update" allowed-tools: WebFetch, Bash metadata: openclaw: emoji: "🧭" homepage: https://github.com/skillnav-dev/skillnav-skill requires: bins: - curl

Route based on $ARGUMENTS[0]:

If $ARGUMENTS is empty or does not match any command above, show this usage message and STOP (do NOT fetch any URL):

SkillNav — AI 开发者工具站 (skillnav.dev)

  /skillnav brief              今日 AI 日报
  /skillnav mcp <keyword>      搜索 MCP Server
  /skillnav search <keyword>   搜索文章、工具、概念
  /skillnav trending           本周热门工具
  /skillnav paper <arxiv-id>   查看论文详情
  /skillnav update             更新到最新版本

Format Rules

Version check

After every API response, compare meta.skill_version from the response with this file's version (2.0.0). If the API version is newer, append at the very end of your output:

💡 SkillNav Skill 有新版本可用 — /skillnav update

brief (full or section-filtered)

If section field is absent (full brief), format as:

1. TL;DR: One bold sentence summarizing the headline
2. Headline title with > why_important in blockquote
3. Bulleted highlights, each with editor comment in parentheses
4. If papers array is non-empty, show "论文速递":

- {title} {attitude} {what} > {implication}

1. If is_fallback is true: "(注意：这是 {date} 的日报，今日暂无更新)"
2. Footer: "完整日报 → {url}"

If section is "papers", format papers only:

1. Header: "论文速递 · {date}"
2. For each paper:

{title} {attitude} {what} > {implication} 趋势：{trend} → arXiv: {url}

1. Footer: "完整日报 → {url}"

If section is "news", format highlights only:

1. Header: "行业动态 · {date}"
2. Headline (if non-empty): {headline.title} — {headline.why_important}
3. Bulleted highlights with editor comments
4. Footer: "完整日报 → {url}"

mcp

Format each result as a numbered list with one blank line between items:

1. {name} {category} ⭐ {stars}

{description_zh} > {editor_comment_zh}

   {install_command}

If editor_comment_zh is null, skip the blockquote line. If results are empty: "未找到匹配的 MCP Server，试试更宽泛的关键词。" Footer: "共 {returned} 个结果{has_more ? '（更多结果请访问 skillnav.dev）' : ''}"

search

Group results by result_type in this order: concept > mcp > skill > article.

Concept: {term}（{zh}）— {one_liner} → {url}

MCP / Skill: {name} ⭐ {stars} — {description_zh} > {editor_comment_zh}

  {install_command}

If editor_comment_zh is null, skip the blockquote line. If install_command is null, skip the code block.

Article: {title_zh} ({source}, {reading_time}min) {summary_zh} → {url}

If results are empty: "未找到与 "{query}" 相关的内容，试试其他关键词。" Footer: "共 {returned} 个结果"

trending

Format as a ranked list grouped by tool_type:

MCP Server

1. {name} ⭐ {stars} (+{weekly_stars_delta} this week)

> {editor_comment_zh}

Skill

1. {name} ⭐ {stars} (+{weekly_stars_delta} this week)

> {editor_comment_zh}

If a group has no items, omit it entirely.

paper (single by ID)

If has_translation is true:

1. {translation.title_zh} {brief_card.attitude} (if brief_card exists)
2. If brief_card exists: {brief_card.what}

> {brief_card.implication}

1. "已有完整中文翻译 → {translation.url}"

If has_translation is false:

1. If brief_card exists:

{brief_card.title} {brief_card.attitude} {brief_card.what} > {brief_card.implication} 趋势：{brief_card.trend}

1. If no brief_card: "未找到该论文的导读信息。"
2. "arXiv 原文 → {arxiv_url}"

paper (keyword search)

1. Header: "论文搜索 · 最近 {days} 天 · "{query}""
2. For each result:

{title} {attitude} · {date} {what} > {implication} → arXiv: {url}

1. If results are empty: "最近 {days} 天未找到与 "{query}" 相关的论文。"
2. Footer: "共 {returned} 篇结果"

Global Footer

Append this line at the very end of EVERY output (after all content, after version check if present):

— SkillNav · skillnav.dev

Error Handling

• If WebFetch fails or returns non-JSON: "SkillNav API 暂时不可用，请直接访问 skillnav.dev"
• If API returns { "error": ... }: Show the message field to the user
• NEVER fabricate data or suggest keywords that weren't in the response