What are Claude Code Skills?

Claude Code Skills are reusable prompt templates stored as markdown files in your project's .claude/skills/ directory. They let you codify best practices and common workflows into slash commands that Claude follows consistently.

How do I create a Claude Code Skill?

Create a markdown file in .claude/skills/ with instructions for Claude. The filename becomes the command name — for example, .claude/skills/code-review.md can be invoked with /code-review during a Claude Code session.

Can I share skills with my team?

Yes. Skills stored in .claude/skills/ can be committed to version control so every team member has access to the same standardized workflows. You can also organize skills into subdirectories by category.

What is the difference between project and global skills?

Project skills (.claude/skills/) are specific to the current project and are committed to git. Global skills (~/.claude/skills/) are personal skills available across all your projects.

smart-search

New

4Community RegistryGeneralby QLBQLB

Automatically route search queries to the optimal MCP engine (Exa, Metaso, Bocha, GitHub, Zai MCP, Fetch) based on query type and content analysis

Community PluginView Source

Overview

Smart Search - Intelligent Search Routing

Automatically route queries to the most appropriate search service using keyword-based priority rules.

Quick Reference

Priority	Service	Trigger Keywords	Token/Count
P0	Exa	code, API, function, framework, library, tutorial, syntax	5000
P1	Metaso	paper, research, study, journal, thesis, academic	10
P2	Brave	today, this week, latest, recent, news	10
P3	Bocha	中文, 国内, 百科, Chinese content	18
P4	GitHub	github, repo, issue, PR, commit, branch	10
P5	Zai MCP	image, video, screenshot, OCR, analyze picture	N/A
P6	Fetch	http://, https:// (direct URL)	5000
P7	Brave	(default fallback)	10
P8	Exa	English technical docs	3000

Decision Flow

code

User Input
    │
    ├─ Contains http:// or https:// ──────────────→ Fetch (P6)
    │
    ├─ Image/Video/Screenshot/OCR ─────────────────→ Zai MCP (P5)
    │
    ├─ GitHub/Repo/Issue/PR/Commit ────────────────→ GitHub MCP (P4)
    │
    ├─ Academic/Paper/Research/Thesis ──────────────→ Metaso (P1)
    │
    ├─ Code/API/Framework/Library ──────────────────→ Exa (P0)
    │
    ├─ Today/Latest/News + "AI" ───────────────────→ Brave + Metaso + Top 8 digest
    │
    ├─ 中文/国内 ───────────────────────────────────→ Bocha (P3)
    │
    └─ Default ─────────────────────────────────────→ Brave (P7)

Token Control

Adjust search depth based on query complexity:

Level	Tokens/Count	Use Case
Quick	5-10	Simple queries, definitions
Regular	10-20	Common usage, examples (default)
Deep	30-50	Multi-condition, architecture

News Output Format (AI新闻输出格式)

When user searches for daily AI news (今日/当天 AI新闻/资讯), use Brave + Metaso combined and output:

code

# 今日AI资讯 Top 8

### 1. [新闻标题]

**摘要：** [新闻内容摘要，50-100字]

**来源：** [来源名](链接)

**标签：** #标签1 #标签2 #标签3

**相关性：** ★★★★★

---

### 2. [新闻标题]

**摘要：** ...

Timeliness Judgment (时效性判断)

Before including news, verify timeliness:

Indicator	Fresh	Stale
Publish date	Within 2 days	Older than 7 days
Date string in URL	Contains current date (2026-02-*)	No date or old date
Source freshness	News sites, blogs	Static docs
Content keywords	"今日", "最新", "just announced"	"历史", "回顾", "总览"

Priority: Prefer news with publish dates within 2 days. Stale news (>7 days) should be excluded.

Source attribution: Always include clickable source links.

Usage Guidelines

DO

•Analyze keywords before routing
•Implement fallback on failure (30s timeout)
•For AI news: Use Brave + Metaso combined
•Format each news item: 标题 + 摘要 + 来源 + 标签 + 相关性
•Filter to recent 2 days: exclude news older than 48 hours
•Always include clickable source links

DON'T

•Use Exa for news → Brave
•Use Metaso for code → Exa
•Include news older than 2 days in "今日" digest
•Omit source attribution for news
•Combine MCPs redundantly

Fallback Strategy

Primary	Fallback 1	Fallback 2
Exa	Brave	Metaso
Metaso	Bocha	Brave
Brave	Bocha	Exa
Bocha	Brave	Exa
GitHub	Fetch	-
Fetch	Brave	-

Combination Rules

Only for complementary scenarios (max 3 concurrent):

Scenario	Combo	Rationale
AI News Digest	Brave + Metaso	English + Chinese comprehensive coverage
Technical research	Exa + GitHub	Code + repos
Deep research	Metaso + Fetch	Academic + content
Comprehensive	Brave + Bocha	News + Chinese

Install & Usage

Create the skills directory

mkdir -p .claude/skills

Download the skill file

mkdir -p .claude/skills && curl -o .claude/skills/smart-search.md https://raw.githubusercontent.com/QLBQLB/smart-search-skill/main/SKILL.md

Invoke in Claude Code

/smart-search

View source on GitHub

mcp

Frequently Asked Questions

What is smart-search?

Automatically route search queries to the optimal MCP engine (Exa, Metaso, Bocha, GitHub, Zai MCP, Fetch) based on query type and content analysis

How to install smart-search?

To install smart-search, create the .claude/skills directory in your project, then run the curl command to download the skill file. Once installed, invoke it in Claude Code with /smart-search.

What is smart-search best for?

smart-search is a community categorized under General. It is designed for: mcp. Created by QLBQLB.