grok-native-search-mcp
Grok Native Search MCP
Ein minimalistischer stdio-MCP-Server mit drei schreibgeschützten Tools:
web_search(query): Grok Native Web Searchx_search(query): Grok Native X Searchweb_fetch(url): Jina Reader: Webseite in Markdown konvertieren
Die Suche verwendet fest grok-4.6, reasoning.effort: low und max_turns: 1. Innerhalb einer Runde dürfen komplementäre Suchen parallel ausgeführt werden; sie stoppen, sobald ausreichend Primärbelege gefunden sind, und erweitern den Umfang nur, wenn Belege fehlen oder widersprüchlich sind. Tool-Antworten werden nicht geparst oder umgeschrieben, sondern direkt als MCP-Text zurückgegeben.
Tool-Routing: Bei bekannten URLs oder wenn der Seiteninhalt verifiziert werden muss, wird web_fetch verwendet; für X-Beiträge, Konten, Threads und Trends wird x_search verwendet; für andere Websuchen ohne eindeutige URL wird web_search verwendet. Nachdem die Suche eine zu verifizierende externe URL geliefert hat, wird anschließend web_fetch aufgerufen.
Umgebungsvariablen
Variable | Zweck |
| API-Key für xAI oder kompatible Gateways |
| Optional, Basisadresse der Responses API; Standard: |
| Jina Reader API-Key |
| Optional, Umgebungs-Proxy-Konfiguration ausschließlich für Jina-Reader-Anfragen |
Related MCP server: Ollama Web Search
Start mit npx
npx -y grok-native-search-mcp@latestCodex-Konfiguration
[mcp_servers.grok_native_search]
command = "npx"
args = ["-y", "grok-native-search-mcp@latest"]
env_vars = ["XAI_API_KEY", "JINA_API_KEY", "HTTP_PROXY", "HTTPS_PROXY", "NO_PROXY"]Starten Sie Codex nach der Konfiguration neu.
Lokale Entwicklung
npm install
npm testAvailable Tools
3 toolsweb_fetchARead-only
Use this when an exact HTTP(S) URL is known, including URLs returned by web_search or x_search, and the task is to read, quote, summarize, or verify page text. Prefer it over Browser, curl, or shell when no login, click, form interaction, or visual inspection is required. Do not use it to discover URLs. Returns raw Markdown from Jina Reader.
Inputs: url (required).
web_searchARead-only
Use this when no exact URL is known and current information must be discovered across the public web. Complementary queries may run in parallel when multiple evidence lanes are needed. Stop after an authoritative source answers the question; deepen only for missing or conflicting evidence. For X-only content use x_search. After finding a page whose contents must be verified, call web_fetch with its exact URL. Returns the raw xAI Responses API body.
Inputs: query (required).
x_searchARead-only
Use this for current X posts, accounts, threads, replies, and trends. Prefer it over web_search for X content. Complementary X queries may run in parallel. Stop when the requested official post or account result is found; deepen only if evidence is missing or conflicting. If a post links to an external page that must be read, call web_fetch with that URL. Returns the raw xAI Responses API body.
Inputs: query (required).
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 3 tools
Each tool has a clearly distinct purpose: web_search for general web discovery, x_search specifically for X content, and web_fetch for retrieving a known URL. The descriptions explicitly cross-reference each other (e.g., web_search notes 'For X-only content use x_search'), leaving no ambiguity about which tool to select.
All three tool names follow a consistent `[source]_[action]` pattern: web_search, x_search, web_fetch. The naming convention is uniform, predictable, and immediately conveys scope (web vs. x) and operation (search vs. fetch).
With only 3 tools, the server is tightly scoped for its purpose: general web search, X-specific search, and URL fetching. This is well within the ideal 3–15 range and every tool earns its place; adding more would likely introduce redundancy.
The surface covers the full lifecycle of a search task: discover sources (web_search, x_search) and then retrieve/verify content (web_fetch). The tools explicitly reference each other to create a complete workflow, with no obvious gaps for the stated domain.
Related MCP Connectors
Live AI-native web search with citations. One tool for every MCP client. Flat per-request pricing.
Official Exa MCP for web search, content fetching, and multi-step research.
Search the agentic web. 4,100+ sites, 11 tools incl. check_url + verify_mcp for probe-before-use.
Enable AI assistants to perform web searches using Perplexity's Sonar Pro.
Related MCP Servers
- AlicenseAqualityDmaintenanceEnables real-time web search via DeepSeek's search-enhanced dialogue, providing search results as answers through MCP tools.1MIT
- AlicenseAqualityBmaintenanceEnables web search and web fetch operations using Ollama's hosted APIs, allowing MCP clients to search the web and retrieve page content.2MIT
- FlicenseNot gradedqualityCmaintenanceEnables web search through one or two SearxNG instances and retrieval of web pages, GitHub READMEs, and article text via MCP tools.-
- AlicenseNot gradedqualityBmaintenanceEnables web search, page fetching and cleaning, OCR from images, and broken-link checking through MCP tools, with multi-provider aggregation, caching, and anti-detection features.35 npmMIT