# Creating a Wikipedia MCP Server in Java in a Few Prompts with Skills > See how I rapidly built a Wikipedia MCP Server in Java through an interactive conversation using Gemini CLI and its powerful skill system. Author: [Guillaume Laforge](https://glaforge.dev/) | Published: 2026-04-02 | Topics: java, jbang, model-context-protocol, ai-agents, langchain4j, gemini-cli, generative-ai, gemini | Canonical: https://glaforge.dev/posts/2026/04/02/creating-a-wikipedia-mcp-server-in-java-in-a-few-prompts/ Since I started using [Model Context Protocol (MCP)](https://modelcontextprotocol.io/) to equip my AI agents with useful tools, I've been looking for ways to quickly build and iterate on local servers. A few weeks ago, I shared [how to easily build a local MCP server in Java with a custom skill in Gemini CLI](https://glaforge.dev/posts/2026/02/21/easily-build-a-local-mcp-server-in-java-with-a-skill-in-gemini-cli/). Today, I wanted to put that skill to the test by creating a **Wikipedia MCP server**. What's impressive is that I didn't even have to leave my terminal or read documentation. The entire process was a conversation with [Gemini CLI](https://geminicli.com), leveraging its ability to search the web, find libraries, and even check migration guides! ## The Interactive Process I started by asking Gemini CLI about the Wikipedia API. Instead of guessing, I used the `@search` command to find the exact _"contracts"_ for searching and retrieving pages. The conversation went something like this: 1. **Exploring the API**: I asked `@search what is the contract for the Wikipedia API to search for Wikipedia pages?`. Gemini found the modern **Wikimedia REST API** (`/search/page`) and the older Action API. ![Using @search in Gemini CLI to find information about the Wikipedia API](/img/gemini-cli/wikipedia-mcp/gcli-mcp-wikipedia-1.jpg) 2. **Retrieving Content**: I then asked how to get the actual page content. It identified the `/page/html/{title}` endpoint as the best way to get clean HTML. ![Gemini CLI screenshot showing the search about the full Wikipedia page retrieval](/img/gemini-cli/wikipedia-mcp/gcli-mcp-wikipedia-2.jpg) 3. **Finding a Converter**: Since LLMs prefer Markdown over raw HTML (returned by the Wikipedia API), I searched for a Java library: `@search how to render HTML to Markdown in Java?`. It suggested **CopyDown** (a Java port of Turndown) as the simplest option. ![Gemini CLI interactive session on how to find an HTML to Markdown converter](/img/gemini-cli/wikipedia-mcp/gcli-mcp-wikipedia-3.jpg) One particularly "pro" move from Gemini: it noticed I wanted to use **Jackson** for JSON parsing. It proactively searched for the latest version, found that **Jackson 3.0.0-rc4** was just released, and even checked the [migration guide](https://github.com/FasterXML/jackson/blob/main/jackson3/MIGRATING_TO_JACKSON_3.md) to ensure the new `tools.jackson` package names were used correctly! Then, I asked Gemini CLI to use my JBang / LangChain4j MCP server creation skill to generate the code of my Wikipedia STDIO MCP server: ![](/img/gemini-cli/wikipedia-mcp/gcli-mcp-wikipedia-4.jpg) ## The Resulting Java Code Once we had the blueprint, it triggered my `jbang-mcp-server` skill. It scaffolded the following [JBang](https://jbang.dev/) script, combining [LangChain4j](https://github.com/langchain4j/langchain4j)'s MCP support with our chosen libraries. ```java ///usr/bin/env jbang "$0" "$@" ; exit $? //DEPS dev.langchain4j:langchain4j-core:1.11.0 //DEPS dev.langchain4j:langchain4j-community-mcp-server:1.11.0-beta19 //DEPS org.slf4j:slf4j-simple:2.0.17 //DEPS io.github.furstenheim:copy_down:1.1 //DEPS tools.jackson.core:jackson-databind:3.0.0-rc4 //JAVA 21 import dev.langchain4j.agent.tool.Tool; import dev.langchain4j.community.mcp.server.McpServer; import dev.langchain4j.community.mcp.server.transport.StdioMcpServerTransport; import org.slf4j.Logger; import org.slf4j.LoggerFactory; import io.github.furstenheim.CopyDown; import tools.jackson.databind.ObjectMapper; import tools.jackson.databind.JsonNode; import tools.jackson.databind.json.JsonMapper; import java.net.URI; import java.net.http.HttpClient; import java.net.http.HttpRequest; import java.net.http.HttpResponse; import java.util.List; import java.util.concurrent.CountDownLatch; import java.net.URLEncoder; import java.nio.charset.StandardCharsets; public class WikipediaMcpServer { static { // Configure SLF4J Simple Logger to write to System.err // This is crucial for MCP servers over STDIO to avoid polluting stdout System.setProperty("org.slf4j.simpleLogger.logFile", "System.err"); } private static final Logger log = LoggerFactory.getLogger(WikipediaMcpServer.class); public static void main(String[] args) throws Exception { log.info("Starting Wikipedia MCP Server..."); WikipediaTools tools = new WikipediaTools(); McpServer server = new McpServer(List.of(tools)); StdioMcpServerTransport transport = new StdioMcpServerTransport(server); log.info("MCP Server started successfully on STDIO."); new CountDownLatch(1).await(); } public static class WikipediaTools { private final HttpClient httpClient = HttpClient.newBuilder().followRedirects(HttpClient.Redirect.NORMAL).build(); private final ObjectMapper objectMapper = JsonMapper.builder().build(); private final CopyDown copyDown = new CopyDown(); private final String USER_AGENT = "WikipediaMcpServer/1.0 (contact@example.com)"; @Tool("Search Wikipedia for a given query and return a list of matching page titles and brief descriptions. Use this to find the exact page title before retrieving content.") public String searchWikipedia(String query) throws Exception { log.info("Searching Wikipedia for: {}", query); String encodedQuery = URLEncoder.encode(query, StandardCharsets.UTF_8); String url = "https://en.wikipedia.org/w/rest.php/v1/search/page?q=" + encodedQuery + "&limit=5"; HttpRequest request = HttpRequest.newBuilder() .uri(URI.create(url)) .header("User-Agent", USER_AGENT) .GET() .build(); HttpResponse response = httpClient.send(request, HttpResponse.BodyHandlers.ofString()); if (response.statusCode() != 200) { return "Error searching Wikipedia: " + response.statusCode() + " - " + response.body(); } JsonNode rootNode = objectMapper.readTree(response.body()); JsonNode pagesNode = rootNode.get("pages"); StringBuilder result = new StringBuilder("Search Results:\n\n"); if (pagesNode != null && pagesNode.isArray()) { for (JsonNode page : pagesNode) { String title = page.has("title") ? page.get("title").asText() : ""; String description = page.has("description") ? page.get("description").asText() : "No description"; result.append("- **").append(title).append("**: ").append(description).append("\n"); } } if (result.length() == "Search Results:\n\n".length()) { return "No results found for query: " + query; } return result.toString(); } @Tool("Retrieve the content of a specific Wikipedia page by its exact title, converted to Markdown format. Use the exact title returned by searchWikipedia.") public String getWikipediaPageContent(String title) throws Exception { log.info("Retrieving Wikipedia page: {}", title); String encodedTitle = URLEncoder.encode(title.replace(" ", "_"), StandardCharsets.UTF_8); String url = "https://en.wikipedia.org/api/rest_v1/page/html/" + encodedTitle; HttpRequest request = HttpRequest.newBuilder() .uri(URI.create(url)) .header("User-Agent", USER_AGENT) .GET() .build(); HttpResponse response = httpClient.send(request, HttpResponse.BodyHandlers.ofString()); if (response.statusCode() != 200) { return "Error retrieving Wikipedia page: " + response.statusCode() + " - " + response.body(); } String htmlContent = response.body(); log.info("Converting HTML to Markdown for page: {}", title); return copyDown.convert(htmlContent); } } } ``` I didn't even touch the code at all, and it worked flawlessly out of the box. I would probably only update the version dependency on LangChain4j which is not up-to-date. But that's about it. ## Configuring the MCP Server in Gemini CLI To use this server, you just need to register it in your `~/.gemini/settings.json`. Gemini CLI will then automatically launch it as a child process and communicate with it over Standard Input/Output. Add the following to your `mcpServers` section: ```json { "mcpServers": { "wikipedia-mcp": { "command": "jbang", "args": [ "run", "--quiet", "/path/to/your/WikipediaMcpServer.java" ] } } } ``` Once the MCP configuration is saved and Gemini CLI reloaded, I could check that the MCP server was available, by running the `/mcp list` command: ![List of available MCP servers tools in Gemini CLI](/img/gemini-cli/wikipedia-mcp/gcli-mcp-wikipedia-5.jpg) There are 2 tools available: one for fetching a list of relevant pages, and the other to fetch the content of an individual page. ## Putting it up to the test Yesterday, the :rocket: Artemis 2 mission launched to travel around the moon :moon: I'm sure Wikipedia is already updated with the latest information about the status of the mission. Let's double check: ![Asking about the Artemis 2 mission](/img/gemini-cli/wikipedia-mcp/gcli-mcp-wikipedia-6.jpg) It found relevant pages, then it loaded the content of those pages and found the information, and summarized its findings: ![Response about the status of the Artemis 2 mission](/img/gemini-cli/wikipedia-mcp/gcli-mcp-wikipedia-7.jpg) Mission in progress! But my own personal mission over those few interactive prompts inside Gemini CLI is accomplished: in less than 5 minutes, I had my custom MCP server to query Wikipedia! And it took me actually more time to write this article itself! ## Wrapping up My winning combo of the day: [Gemini CLI](https://geminicli.com), agent skills, Java, [JBang](https://www.jbang.dev/), and [LangChain4j](https://docs.langchain4j.dev)... and boom :boom: Maybe we'll find that obvious in a few months, but I'm still impressed by how it is today, in just a few prompts, to create something useful like an MCP server that your friendly AI agents can use. Building tools for your AI agents has never been this fluid. Happy hacking! 🚀 And let's go to the :moon: and beyond!