메뉴
Perform an AI-native web search using the official Tavily MCP server. Use when the user needs high-quality technical search results, coding documentation, or token-efficient AI-optimized search.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
SQL engineering standards for writing correct, safe, cross-dialect SQL across PostgreSQL, MySQL, SQL Server, SQLite, DuckDB, and Oracle. Use when writing SQL queries, exploring schemas, running migrations, optimizing queries, transpiling between dialects, or connecting to databases via native CLIs.
MongoDB engineering standards for querying collections, building aggregation pipelines, designing document schemas, managing indexes, and administering deployments via mongosh. Use when writing MongoDB queries, exploring collections, designing schemas, optimizing performance, or managing replica sets and sharding.
Engineering planning standards for creating design docs, DRI working docs, ADRs, and weekly updates. Use when planning projects, writing design docs, creating RFCs, tracking risks, writing status updates, or reviewing engineering plans. Covers structure, operational tracking, context engineering, and scaling from startup to enterprise.
Multi-agent PR review that spawns parallel domain-specific subagents (python-expert, rust-expert, dbt-expert, dotnet-expert, kedro-expert, d2-tala-expert, ui-designer) based on changed file types, posts inline review comments on GitHub, GitLab, and Azure DevOps.
dbt engineering standards for building clean, modular, well-tested data transformations. Use when writing dbt models, configuring sources, designing project structure, writing tests, managing materializations, or preparing for dbt Fusion. Covers dbt Core 1.8–1.11, Fusion v2.0, SQL style, testing, governance, and deployment.
Langfuse observability platform standards for querying traces, managing prompts, debugging LLM applications, and accessing Langfuse data programmatically. Use when interacting with Langfuse, querying traces or sessions, managing prompts, instrumenting applications, or looking up Langfuse documentation. Covers CLI-based API access (via npx) and documentation retrieval.
| name | tavily-search |
| description | Perform an AI-native web search using the official Tavily MCP server. Use when the user needs high-quality technical search results, coding documentation, or token-efficient AI-optimized search. |
| argument-hint | <query> |
Perform an AI-native web search using the official Tavily MCP server.
IMPORTANT: You MUST use the mcp__tavily__tavily_search tool for this command. Do NOT use the built-in WebSearch tool or Brave Search MCP tools.
.mcp.json using tavily-mcpTAVILY_API_KEY environment variable set (get one at https://tavily.com)query: The search terms (everything after /tavily-search)When invoked:
mcp__tavily__tavily_search call with the provided query — never fire multiple searches per invocation/tavily-search latest Next.js features
/tavily-search how to configure ESLint for TypeScript
/tavily-search React Server Components best practices
## Tavily Search: [query]
### Results
[Summarized findings from search results]
### Sources
- [Title](URL)
- [Title](URL)
- [Title](URL)
If the search fails:
TAVILY_API_KEY is set correctly/mcp to verify status)/brave-search or /web-search as a fallback/brave-search - Uses Brave Search MCP (best for image, video, news, local search)/web-search - Uses Claude's built-in WebSearch (no MCP required)internet-researcher agent - For deep, multi-query research