mcp-searxng 1.3.4 → 1.5.0

package/README.md

@@ -40,6 +40,10 @@ Replace `YOUR_SEARXNG_INSTANCE_URL` with the URL of your SearXNG instance (e.g.
 ## Features
 
 - **Web Search**: General queries, news, articles, with pagination.
+- **Structured Search Output**: Choose formatted text or raw SearXNG-shaped JSON with `response_format`.
+- **Direct Answers & Metadata**: Text results surface SearXNG answers, corrections, suggestions, and infoboxes before result lists.
+- **Search Suggestions**: Query autocomplete via SearXNG's `/autocompleter` endpoint.
+- **Instance Capability Discovery**: Inspect configured categories, engines, defaults, locales, and plugins from `/config`.
 - **URL Content Reading**: Advanced content extraction with pagination, section filtering, and heading extraction.
 - **Intelligent Caching**: URL content is cached with TTL (Time-To-Live) to improve performance and reduce redundant requests.
 - **Pagination**: Control which page of results to retrieve.
@@ -86,6 +90,23 @@ AI Assistant (e.g. Claude)
     - `language` (string, optional): Language code for results (e.g., "en", "fr", "de") or "all" (default: "all")
     - `safesearch` (number, optional): Safe search filter level (0: None, 1: Moderate, 2: Strict) (default: instance setting)
     - `min_score` (number, optional): Minimum relevance score from 0.0 to 1.0. Results below this score are filtered out.
+    - `num_results` (number, optional): Maximum number of results to return, from 1 to 20. `SEARXNG_MAX_RESULTS` applies as an operator ceiling.
+    - `categories` (string, optional): Comma-separated SearXNG categories (e.g. `"news"`, `"it,science"`). Supported values: `general`, `news`, `images`, `videos`, `it`, `science`, `files`, `social media`. Default: SearXNG instance default (usually `general`).
+    - `response_format` (string, optional): Response format, either `"text"` for formatted agent-readable output or `"json"` for raw SearXNG JSON with filtered/sliced `results`. (default: `"text"`)
+
+- **searxng_search_suggestions**
+  - Get autocomplete suggestions for refining search queries
+  - Inputs:
+    - `query` (string): Partial or complete query to autocomplete.
+    - `language` (string, optional): Language code for suggestions (e.g., "en", "fr", "de") or "all" (default: "all")
+
+- **searxng_instance_info**
+  - Discover categories, engines, defaults, locales, and plugins exposed by the configured SearXNG instance
+  - Inputs:
+    - `includeEngines` (boolean, optional): Include enabled engine names in the response. (default: false)
+    - `includeDisabled` (boolean, optional): Include disabled engine names when `includeEngines` is true. (default: false)
+    - `category` (string, optional): Filter categories and engines to a single category name.
+    - `refresh` (boolean, optional): Bypass the process cache and fetch fresh `/config` data. (default: false)
 
 - **web_url_read**
   - Read and convert the content from a URL to markdown with advanced content extraction options

package/dist/index.js

@@ -2,9 +2,11 @@ import { McpServer } from "@modelcontextprotocol/sdk/server/mcp.js";
 import { StdioServerTransport } from "@modelcontextprotocol/sdk/server/stdio.js";
 import { CallToolRequestSchema, ListToolsRequestSchema, SetLevelRequestSchema, ListResourcesRequestSchema, ListResourceTemplatesRequestSchema, ReadResourceRequestSchema, } from "@modelcontextprotocol/sdk/types.js";
 // Import modularized functionality
-import { WEB_SEARCH_TOOL, READ_URL_TOOL, isSearXNGWebSearchArgs } from "./types.js";
+import { WEB_SEARCH_TOOL, SUGGESTIONS_TOOL, INSTANCE_INFO_TOOL, READ_URL_TOOL, LITE_WEB_SEARCH_TOOL, LITE_SUGGESTIONS_TOOL, LITE_INSTANCE_INFO_TOOL, LITE_READ_URL_TOOL, isSearXNGWebSearchArgs, isSearXNGSearchSuggestionsArgs, isSearXNGInstanceInfoArgs, } from "./types.js";
 import { logMessage, setLogLevel, getCurrentLogLevel } from "./logging.js";
 import { performWebSearch } from "./search.js";
+import { performSearchSuggestions } from "./suggestions.js";
+import { fetchInstanceInfo } from "./instance-info.js";
 import { fetchAndConvertToMarkdown } from "./url-reader.js";
 import { createConfigResource, createHelpResource } from "./resources.js";
 import { createHttpServer, resolveBindHost } from "./http-server.js";
@@ -41,6 +43,30 @@ export function isWebUrlReadArgs(args) {
     }
     return true;
 }
+function getFetchTimeoutMs(mcpServer) {
+    const rawValue = process.env.FETCH_TIMEOUT_MS;
+    if (rawValue === undefined || rawValue.trim() === "") {
+        return 10000;
+    }
+    const parsed = parseInt(rawValue, 10);
+    if (Number.isNaN(parsed) || parsed <= 0) {
+        logMessage(mcpServer, "warning", `Ignoring invalid FETCH_TIMEOUT_MS="${rawValue}". Expected a positive integer. Using default 10000.`);
+        return 10000;
+    }
+    return parsed;
+}
+function getDefaultUrlReadMaxChars(mcpServer) {
+    const rawValue = process.env.URL_READ_MAX_CHARS;
+    if (rawValue === undefined || rawValue.trim() === "") {
+        return undefined;
+    }
+    const parsed = parseInt(rawValue, 10);
+    if (Number.isNaN(parsed) || parsed <= 0) {
+        logMessage(mcpServer, "warning", `Ignoring invalid URL_READ_MAX_CHARS="${rawValue}". Expected a positive integer.`);
+        return undefined;
+    }
+    return parsed;
+}
 /**
  * Creates and configures a new McpServer with all handlers registered.
  * Called once per HTTP session, or once for STDIO mode.
@@ -57,11 +83,16 @@ export function createMcpServer() {
         },
     });
     const server = mcpServer.server;
+    const useLiteTools = process.env.SEARXNG_LITE_TOOLS === "true";
+    const searchTool = useLiteTools ? LITE_WEB_SEARCH_TOOL : WEB_SEARCH_TOOL;
+    const suggestionsTool = useLiteTools ? LITE_SUGGESTIONS_TOOL : SUGGESTIONS_TOOL;
+    const instanceInfoTool = useLiteTools ? LITE_INSTANCE_INFO_TOOL : INSTANCE_INFO_TOOL;
+    const readUrlTool = useLiteTools ? LITE_READ_URL_TOOL : READ_URL_TOOL;
     // List tools handler
     server.setRequestHandler(ListToolsRequestSchema, async () => {
         logMessage(mcpServer, "debug", "Handling list_tools request");
         return {
-            tools: [WEB_SEARCH_TOOL, READ_URL_TOOL],
+            tools: [searchTool, suggestionsTool, instanceInfoTool, readUrlTool],
         };
     });
     // Call tool handler
@@ -73,7 +104,35 @@ export function createMcpServer() {
                 if (!isSearXNGWebSearchArgs(args)) {
                     throw new Error("Invalid arguments for web search");
                 }
-                const result = await performWebSearch(mcpServer, args.query, args.pageno, args.time_range, args.language, args.safesearch, args.min_score);
+                const result = await performWebSearch(mcpServer, args.query, args.pageno, args.time_range, args.language, args.safesearch, args.min_score, args.num_results, args.categories, args.response_format);
+                return {
+                    content: [
+                        {
+                            type: "text",
+                            text: result,
+                        },
+                    ],
+                };
+            }
+            else if (name === "searxng_search_suggestions") {
+                if (!isSearXNGSearchSuggestionsArgs(args)) {
+                    throw new Error("Invalid arguments for search suggestions");
+                }
+                const suggestions = await performSearchSuggestions(mcpServer, args.query, args.language);
+                return {
+                    content: [
+                        {
+                            type: "text",
+                            text: JSON.stringify({ query: args.query, suggestions }, null, 2),
+                        },
+                    ],
+                };
+            }
+            else if (name === "searxng_instance_info") {
+                if (!isSearXNGInstanceInfoArgs(args)) {
+                    throw new Error("Invalid arguments for instance info");
+                }
+                const result = await fetchInstanceInfo(mcpServer, args.includeEngines, args.includeDisabled, args.category, args.refresh);
                 return {
                     content: [
                         {
@@ -87,14 +146,15 @@ export function createMcpServer() {
                 if (!isWebUrlReadArgs(args)) {
                     throw new Error("Invalid arguments for URL reading");
                 }
+                const defaultMaxLength = getDefaultUrlReadMaxChars(mcpServer);
                 const paginationOptions = {
                     startChar: args.startChar,
-                    maxLength: args.maxLength,
+                    maxLength: args.maxLength ?? defaultMaxLength,
                     section: args.section,
                     paragraphRange: args.paragraphRange,
                     readHeadings: args.readHeadings,
                 };
-                const result = await fetchAndConvertToMarkdown(mcpServer, args.url, 10000, paginationOptions);
+                const result = await fetchAndConvertToMarkdown(mcpServer, args.url, getFetchTimeoutMs(mcpServer), paginationOptions);
                 return {
                     content: [
                         {

package/dist/instance-info.d.ts

@@ -0,0 +1,3 @@
+import { McpServer } from "@modelcontextprotocol/sdk/server/mcp.js";
+export declare function clearInstanceInfoCacheForTests(): void;
+export declare function fetchInstanceInfo(mcpServer: McpServer, includeEngines?: boolean, includeDisabled?: boolean, category?: string, refresh?: boolean): Promise<string>;

package/dist/instance-info.js

@@ -0,0 +1,113 @@
+import { logMessage } from "./logging.js";
+import { createDefaultAgent, createProxyAgent, ProxyType } from "./proxy.js";
+let cachedConfig = null;
+let cachedBaseUrl = null;
+function unavailable(message, status) {
+    return JSON.stringify({
+        available: false,
+        message,
+        ...(status !== undefined ? { status } : {}),
+    }, null, 2);
+}
+function namesFromCategories(config) {
+    if (!config.categories || typeof config.categories !== "object") {
+        return [];
+    }
+    return Object.keys(config.categories).sort();
+}
+function engineCategories(engine) {
+    if (Array.isArray(engine.categories)) {
+        return engine.categories;
+    }
+    if (typeof engine.category === "string") {
+        return [engine.category];
+    }
+    return [];
+}
+function collectEngines(config, includeDisabled, category) {
+    const enabled = new Set();
+    const disabled = new Set();
+    if (Array.isArray(config.engines)) {
+        for (const engine of config.engines) {
+            if (!engine || typeof engine.name !== "string") {
+                continue;
+            }
+            const categories = engineCategories(engine);
+            if (category && !categories.includes(category)) {
+                continue;
+            }
+            if (engine.disabled) {
+                if (includeDisabled) {
+                    disabled.add(engine.name);
+                }
+            }
+            else {
+                enabled.add(engine.name);
+            }
+        }
+    }
+    return {
+        enabled: [...enabled].sort(),
+        ...(includeDisabled ? { disabled: [...disabled].sort() } : {}),
+    };
+}
+function formatInstanceInfo(config, includeEngines, includeDisabled, category) {
+    const categories = category
+        ? namesFromCategories(config).filter((name) => name === category)
+        : namesFromCategories(config);
+    const payload = {
+        available: true,
+        categories,
+        defaults: {
+            safesearch: config.search?.safe_search ?? config.default_safe_search,
+            locale: config.default_locale,
+            language: config.default_language,
+            theme: config.default_theme,
+        },
+        locales: config.locales,
+        plugins: config.plugins ?? [],
+    };
+    if (includeEngines) {
+        payload.engines = collectEngines(config, includeDisabled, category);
+    }
+    return JSON.stringify(payload, null, 2);
+}
+export function clearInstanceInfoCacheForTests() {
+    cachedConfig = null;
+    cachedBaseUrl = null;
+}
+export async function fetchInstanceInfo(mcpServer, includeEngines = false, includeDisabled = false, category, refresh = false) {
+    const base = process.env.SEARXNG_URL;
+    if (!base) {
+        return unavailable("SEARXNG_URL is not configured; cannot fetch SearXNG /config.");
+    }
+    if (refresh) {
+        cachedConfig = null;
+    }
+    if (cachedConfig && cachedBaseUrl === base) {
+        return formatInstanceInfo(cachedConfig, includeEngines, includeDisabled, category);
+    }
+    const parsedBase = new URL(base.endsWith("/") ? base : `${base}/`);
+    const url = new URL("config", parsedBase);
+    try {
+        const requestOptions = {
+            signal: AbortSignal.timeout(5000),
+        };
+        const proxyAgent = createProxyAgent(url.toString(), ProxyType.SEARCH);
+        const dispatcher = proxyAgent ?? createDefaultAgent();
+        if (dispatcher) {
+            requestOptions.dispatcher = dispatcher;
+        }
+        const response = await fetch(url.toString(), requestOptions);
+        if (!response.ok) {
+            return unavailable(`SearXNG /config is unavailable: HTTP ${response.status} ${response.statusText}`, response.status);
+        }
+        cachedConfig = await response.json();
+        cachedBaseUrl = base;
+        return formatInstanceInfo(cachedConfig, includeEngines, includeDisabled, category);
+    }
+    catch (error) {
+        logMessage(mcpServer, "warning", `SearXNG /config fetch failed: ${error instanceof Error ? error.message : String(error)}`);
+        return unavailable("SearXNG /config is unavailable; instance capability discovery could not complete.");
+    }
+}

package/dist/resources.js

@@ -21,7 +21,12 @@ export function createConfigResource() {
             currentLogLevel: getCurrentLogLevel()
         },
         capabilities: {
-            tools: ["searxng_web_search", "web_url_read"],
+            tools: [
+                "searxng_web_search",
+                "searxng_search_suggestions",
+                "searxng_instance_info",
+                "web_url_read"
+            ],
             logging: true,
             resources: true,
             transports: process.env.MCP_HTTP_PORT ? ["stdio", "http"] : ["stdio"]
@@ -33,7 +38,7 @@ export function createHelpResource() {
     return `# SearXNG MCP Server Help
 
 ## Overview
-This is a Model Context Protocol (MCP) server that provides web search capabilities through SearXNG and URL content reading functionality.
+This is a Model Context Protocol (MCP) server that provides web search, autocomplete suggestions, instance capability discovery, and URL content reading through SearXNG.
 
 ## Available Tools
 
@@ -47,12 +52,38 @@ Performs web searches using the configured SearXNG instance.
 - \`language\` (optional): Language code like "en", "fr", "de" (default: "all")
 - \`safesearch\` (optional): Safe search level - 0 (none), 1 (moderate), 2 (strict)
 - \`min_score\` (optional): Minimum relevance score from 0.0 to 1.0
+- \`num_results\` (optional): Maximum result count from 1 to 20
+- \`categories\` (optional): Comma-separated SearXNG categories such as "news" or "it,science"
+- \`response_format\` (optional): "text" for formatted output or "json" for raw SearXNG-shaped JSON
 
-### 2. web_url_read
+Text output can include metadata sections for direct answers, spelling corrections, suggestions, and infoboxes before the result list. JSON output preserves the SearXNG response shape with filtered and sliced \`results\`.
+
+### 2. searxng_search_suggestions
+Returns autocomplete suggestions from the configured SearXNG instance.
+
+**Parameters:**
+- \`query\` (required): Partial or complete query to autocomplete
+- \`language\` (optional): Language code like "en", "fr", "de" or "all" (default: "all")
+
+### 3. searxng_instance_info
+Discovers categories, engines, defaults, locales, and plugins exposed by the configured SearXNG instance.
+
+**Parameters:**
+- \`includeEngines\` (optional): Include enabled engine names
+- \`includeDisabled\` (optional): Include disabled engine names when \`includeEngines\` is true
+- \`category\` (optional): Filter categories and engines to one category
+- \`refresh\` (optional): Bypass the process cache and fetch fresh \`/config\` data
+
+### 4. web_url_read
 Reads and converts web page content to Markdown format.
 
 **Parameters:**
 - \`url\` (required): The URL to fetch and convert
+- \`startChar\` (optional): Starting character position
+- \`maxLength\` (optional): Maximum number of characters to return
+- \`section\` (optional): Extract content under a heading
+- \`paragraphRange\` (optional): Return a paragraph range such as "1-5" or "10-"
+- \`readHeadings\` (optional): Return only headings
 
 ## Configuration
 
@@ -98,6 +129,18 @@ Tool: web_url_read
 Args: {"url": "https://example.com/article"}
 \`\`\`
 
+### Get query suggestions
+\`\`\`
+Tool: searxng_search_suggestions
+Args: {"query": "typescr"}
+\`\`\`
+
+### Discover instance capabilities
+\`\`\`
+Tool: searxng_instance_info
+Args: {"includeEngines": true}
+\`\`\`
+
 ## Troubleshooting
 
 1. **"SEARXNG_URL not set"**: Configure the SEARXNG_URL environment variable

package/dist/search.d.ts

@@ -1,2 +1,2 @@
 import { McpServer } from "@modelcontextprotocol/sdk/server/mcp.js";
-export declare function performWebSearch(mcpServer: McpServer, query: string, pageno?: number, time_range?: string, language?: string, safesearch?: number, min_score?: number): Promise<string>;
+export declare function performWebSearch(mcpServer: McpServer, query: string, pageno?: number, time_range?: string, language?: string, safesearch?: number, min_score?: number, num_results?: number, categories?: string, response_format?: "text" | "json"): Promise<string>;

package/dist/search.js

@@ -1,14 +1,100 @@
 import { createProxyAgent, createDefaultAgent, ProxyType } from "./proxy.js";
 import { logMessage } from "./logging.js";
 import { MCPSearXNGError, validateEnvironment, createNetworkError, createServerError, createJSONError, createDataError, createNoResultsMessage } from "./error-handler.js";
-export async function performWebSearch(mcpServer, query, pageno = 1, time_range, language = "all", safesearch, min_score) {
+function getOperatorMaxResults(mcpServer) {
+    const rawValue = process.env.SEARXNG_MAX_RESULTS;
+    if (rawValue === undefined || rawValue.trim() === "") {
+        return undefined;
+    }
+    const parsed = parseInt(rawValue, 10);
+    if (Number.isNaN(parsed) || parsed <= 0 || parsed > 20) {
+        logMessage(mcpServer, "warning", `Ignoring invalid SEARXNG_MAX_RESULTS="${rawValue}". Expected an integer from 1 to 20.`);
+        return undefined;
+    }
+    return parsed;
+}
+function getMaxResultChars(mcpServer) {
+    const rawValue = process.env.SEARXNG_MAX_RESULT_CHARS;
+    if (rawValue === undefined || rawValue.trim() === "") {
+        return undefined;
+    }
+    const parsed = parseInt(rawValue, 10);
+    if (Number.isNaN(parsed) || parsed <= 0) {
+        logMessage(mcpServer, "warning", `Ignoring invalid SEARXNG_MAX_RESULT_CHARS="${rawValue}". Expected a positive integer.`);
+        return undefined;
+    }
+    return parsed;
+}
+function truncateResultContent(content, maxResultChars) {
+    if (maxResultChars === undefined || content.length <= maxResultChars) {
+        return content;
+    }
+    return `${content.slice(0, maxResultChars)}…`;
+}
+function hasItems(items) {
+    return Array.isArray(items) && items.length > 0;
+}
+function formatSearchMetadata(data) {
+    const sections = [];
+    if (hasItems(data.answers)) {
+        sections.push(data.answers.map((answer) => `Direct answer: ${answer}`).join("\n"));
+    }
+    if (hasItems(data.corrections)) {
+        sections.push(data.corrections.map((correction) => `Spelling correction: did you mean "${correction}"?`).join("\n"));
+    }
+    if (hasItems(data.suggestions)) {
+        sections.push(`Suggestions: ${data.suggestions.join(", ")}`);
+    }
+    if (hasItems(data.infoboxes)) {
+        const infoboxText = data.infoboxes
+            .map((infobox) => {
+            const lines = [`Infobox: ${infobox.infobox}`];
+            if (infobox.content) {
+                lines.push(infobox.content);
+            }
+            if (hasItems(infobox.urls)) {
+                lines.push(...infobox.urls.map((entry) => `${entry.title}: ${entry.url}`));
+            }
+            return lines.join("\n");
+        })
+            .join("\n\n");
+        sections.push(infoboxText);
+    }
+    return sections.join("\n\n");
+}
+function getDefaultLanguage() {
+    return process.env.SEARXNG_DEFAULT_LANGUAGE ?? "all";
+}
+function getDefaultSafesearch(mcpServer) {
+    const rawValue = process.env.SEARXNG_DEFAULT_SAFESEARCH;
+    if (rawValue === undefined || rawValue.trim() === "") {
+        return undefined;
+    }
+    const parsed = parseInt(rawValue, 10);
+    if (Number.isNaN(parsed) || ![0, 1, 2].includes(parsed)) {
+        logMessage(mcpServer, "warning", `Ignoring invalid SEARXNG_DEFAULT_SAFESEARCH="${rawValue}". Expected 0, 1, or 2.`);
+        return undefined;
+    }
+    return parsed;
+}
+export async function performWebSearch(mcpServer, query, pageno = 1, time_range, language, safesearch, min_score, num_results, categories, response_format = "text") {
     const startTime = Date.now();
+    const operatorMax = getOperatorMaxResults(mcpServer);
+    const effectiveMax = operatorMax !== undefined
+        ? (num_results !== undefined ? Math.min(num_results, operatorMax) : operatorMax)
+        : num_results;
+    const maxResultChars = getMaxResultChars(mcpServer);
+    const effectiveLanguage = language ?? getDefaultLanguage();
+    const effectiveSafesearch = safesearch !== undefined ? safesearch : getDefaultSafesearch(mcpServer);
     // Build detailed log message with all parameters
     const searchParams = [
         `page ${pageno}`,
-        `lang: ${language}`,
+        `lang: ${effectiveLanguage}`,
         time_range ? `time: ${time_range}` : null,
-        safesearch ? `safesearch: ${safesearch}` : null
+        effectiveSafesearch !== undefined ? `safesearch: ${effectiveSafesearch}` : null,
+        min_score !== undefined ? `min_score: ${min_score}` : null,
+        effectiveMax !== undefined ? `num_results: ${effectiveMax}` : null,
+        categories ? `categories: ${categories}` : null,
     ].filter(Boolean).join(", ");
     logMessage(mcpServer, "info", `Starting web search: "${query}" (${searchParams})`);
     const validationError = validateEnvironment();
@@ -26,11 +112,14 @@ export async function performWebSearch(mcpServer, query, pageno = 1, time_range,
         ["day", "week", "month", "year"].includes(time_range)) {
         url.searchParams.set("time_range", time_range);
     }
-    if (language && language !== "all") {
-        url.searchParams.set("language", language);
+    if (effectiveLanguage && effectiveLanguage !== "all") {
+        url.searchParams.set("language", effectiveLanguage);
     }
-    if (safesearch !== undefined && [0, 1, 2].includes(safesearch)) {
-        url.searchParams.set("safesearch", safesearch.toString());
+    if (effectiveSafesearch !== undefined && [0, 1, 2].includes(effectiveSafesearch)) {
+        url.searchParams.set("safesearch", effectiveSafesearch.toString());
+    }
+    if (categories) {
+        url.searchParams.set("categories", categories);
     }
     // Prepare request options with headers
     const requestOptions = {
@@ -119,21 +208,31 @@ export async function performWebSearch(mcpServer, query, pageno = 1, time_range,
         throw createDataError(data, context);
     }
     const results = data.results
-        .map((result) => ({
-        title: result.title || "",
-        content: result.content || "",
-        url: result.url || "",
-        score: result.score || 0,
-    }))
-        .filter((result) => min_score === undefined || result.score >= min_score);
-    if (results.length === 0) {
-        const filterNote = min_score === undefined ? "" : ` after applying min_score=${min_score}`;
+        .filter((result) => min_score === undefined || (result.score || 0) >= min_score);
+    const slicedResults = effectiveMax !== undefined
+        ? results.slice(0, effectiveMax)
+        : results;
+    if (response_format === "json") {
+        return JSON.stringify({ ...data, results: slicedResults }, null, 2);
+    }
+    const metadata = formatSearchMetadata(data);
+    if (slicedResults.length === 0) {
+        const appliedFilters = [
+            min_score === undefined ? null : `min_score=${min_score}`,
+            effectiveMax === undefined ? null : `num_results=${effectiveMax}`,
+        ].filter(Boolean).join(" ");
+        const filterNote = appliedFilters ? ` after applying ${appliedFilters}` : "";
         logMessage(mcpServer, "info", `No results found for query: "${query}"${filterNote}`);
-        return createNoResultsMessage(query);
+        const noResultsMessage = createNoResultsMessage(query);
+        return metadata ? `${metadata}\n\n---\n\n${noResultsMessage}` : noResultsMessage;
     }
     const duration = Date.now() - startTime;
-    logMessage(mcpServer, "info", `Search completed: "${query}" (${searchParams}) - ${results.length} results in ${duration}ms`);
-    return results
-        .map((r) => `Title: ${r.title}\nDescription: ${r.content}\nURL: ${r.url}\nRelevance Score: ${r.score.toFixed(3)}`)
+    logMessage(mcpServer, "info", `Search completed: "${query}" (${searchParams}) - ${slicedResults.length} results in ${duration}ms`);
+    const formattedResults = slicedResults
+        .map((r) => {
+        const score = r.score || 0;
+        return `Title: ${r.title || ""}\nDescription: ${truncateResultContent(r.content || "", maxResultChars)}\nURL: ${r.url || ""}\nRelevance Score: ${score.toFixed(3)}`;
+    })
         .join("\n\n");
+    return metadata ? `${metadata}\n\n---\n\n${formattedResults}` : formattedResults;
 }

package/dist/suggestions.d.ts

@@ -0,0 +1,2 @@
+import { McpServer } from "@modelcontextprotocol/sdk/server/mcp.js";
+export declare function performSearchSuggestions(mcpServer: McpServer, query: string, language?: string): Promise<string[]>;

package/dist/suggestions.js

@@ -0,0 +1,34 @@
+import { logMessage } from "./logging.js";
+import { createDefaultAgent, createProxyAgent, ProxyType } from "./proxy.js";
+export async function performSearchSuggestions(mcpServer, query, language = "all") {
+    const base = process.env.SEARXNG_URL;
+    if (!base) {
+        return [];
+    }
+    const parsedBase = new URL(base.endsWith("/") ? base : `${base}/`);
+    const url = new URL("autocompleter", parsedBase);
+    url.searchParams.set("q", query);
+    if (language !== "all") {
+        url.searchParams.set("lang", language);
+    }
+    try {
+        const requestOptions = {
+            signal: AbortSignal.timeout(5000),
+        };
+        const proxyAgent = createProxyAgent(url.toString(), ProxyType.SEARCH);
+        const dispatcher = proxyAgent ?? createDefaultAgent();
+        if (dispatcher) {
+            requestOptions.dispatcher = dispatcher;
+        }
+        const response = await fetch(url.toString(), requestOptions);
+        if (!response.ok) {
+            return [];
+        }
+        const data = await response.json();
+        return Array.isArray(data[1]) ? data[1] : [];
+    }
+    catch {
+        logMessage(mcpServer, "debug", "Autocomplete request failed; returning empty suggestions");
+        return [];
+    }
+}

package/dist/types.d.ts

@@ -36,6 +36,25 @@ export declare function isSearXNGWebSearchArgs(args: unknown): args is {
     language?: string;
     safesearch?: number;
     min_score?: number;
+    num_results?: number;
+    categories?: string;
+    response_format?: "text" | "json";
+};
+export declare function isSearXNGSearchSuggestionsArgs(args: unknown): args is {
+    query: string;
+    language?: string;
+};
+export declare function isSearXNGInstanceInfoArgs(args: unknown): args is {
+    includeEngines?: boolean;
+    includeDisabled?: boolean;
+    category?: string;
+    refresh?: boolean;
 };
 export declare const WEB_SEARCH_TOOL: Tool;
+export declare const SUGGESTIONS_TOOL: Tool;
+export declare const INSTANCE_INFO_TOOL: Tool;
+export declare const LITE_WEB_SEARCH_TOOL: Tool;
+export declare const LITE_SUGGESTIONS_TOOL: Tool;
+export declare const LITE_INSTANCE_INFO_TOOL: Tool;
+export declare const LITE_READ_URL_TOOL: Tool;
 export declare const READ_URL_TOOL: Tool;

package/dist/types.js

@@ -1,5 +1,6 @@
 const VALID_TIME_RANGES = ["day", "week", "month", "year"];
 const VALID_SAFESEARCH_VALUES = [0, 1, 2];
+const VALID_RESPONSE_FORMATS = ["text", "json"];
 export function isSearXNGWebSearchArgs(args) {
     if (typeof args !== "object" ||
         args === null ||
@@ -29,6 +30,53 @@ export function isSearXNGWebSearchArgs(args) {
             searchArgs.min_score > 1)) {
         return false;
     }
+    if (searchArgs.num_results !== undefined &&
+        (typeof searchArgs.num_results !== "number" ||
+            Number.isNaN(searchArgs.num_results) ||
+            !Number.isInteger(searchArgs.num_results) ||
+            searchArgs.num_results < 1 ||
+            searchArgs.num_results > 20)) {
+        return false;
+    }
+    if (searchArgs.categories !== undefined && typeof searchArgs.categories !== "string") {
+        return false;
+    }
+    if (searchArgs.response_format !== undefined &&
+        (typeof searchArgs.response_format !== "string" || !VALID_RESPONSE_FORMATS.includes(searchArgs.response_format))) {
+        return false;
+    }
+    return true;
+}
+export function isSearXNGSearchSuggestionsArgs(args) {
+    if (typeof args !== "object" ||
+        args === null ||
+        !("query" in args) ||
+        typeof args.query !== "string") {
+        return false;
+    }
+    const suggestionArgs = args;
+    if (suggestionArgs.language !== undefined && typeof suggestionArgs.language !== "string") {
+        return false;
+    }
+    return true;
+}
+export function isSearXNGInstanceInfoArgs(args) {
+    if (typeof args !== "object" || args === null) {
+        return false;
+    }
+    const infoArgs = args;
+    if (infoArgs.includeEngines !== undefined && typeof infoArgs.includeEngines !== "boolean") {
+        return false;
+    }
+    if (infoArgs.includeDisabled !== undefined && typeof infoArgs.includeDisabled !== "boolean") {
+        return false;
+    }
+    if (infoArgs.category !== undefined && typeof infoArgs.category !== "string") {
+        return false;
+    }
+    if (infoArgs.refresh !== undefined && typeof infoArgs.refresh !== "boolean") {
+        return false;
+    }
     return true;
 }
 export const WEB_SEARCH_TOOL = {
@@ -76,10 +124,119 @@ export const WEB_SEARCH_TOOL = {
                 minimum: 0,
                 maximum: 1,
             },
+            num_results: {
+                type: "number",
+                description: "Maximum number of results to return (1-20). Operator cap SEARXNG_MAX_RESULTS applies as a ceiling.",
+                minimum: 1,
+                maximum: 20,
+            },
+            categories: {
+                type: "string",
+                description: "Comma-separated SearXNG categories. Supported: general, news, images, videos, it, science, files, social media. Default: general (SearXNG instance default).",
+            },
+            response_format: {
+                type: "string",
+                description: "Response format: formatted text for agents or raw JSON for programmatic clients. Default: text.",
+                enum: ["text", "json"],
+                default: "text",
+            },
+        },
+        required: ["query"],
+    },
+};
+export const SUGGESTIONS_TOOL = {
+    name: "searxng_search_suggestions",
+    description: "Returns autocomplete suggestions from the configured SearXNG instance. " +
+        "Use this to refine vague or partial queries before searching.",
+    annotations: {
+        readOnlyHint: true,
+        openWorldHint: true,
+    },
+    inputSchema: {
+        type: "object",
+        properties: {
+            query: {
+                type: "string",
+                description: "Partial or complete query to autocomplete.",
+            },
+            language: {
+                type: "string",
+                description: "Language code for suggestions (e.g., 'en', 'fr', 'de') or 'all'. Default: all.",
+                default: "all",
+            },
+        },
+        required: ["query"],
+    },
+};
+export const INSTANCE_INFO_TOOL = {
+    name: "searxng_instance_info",
+    description: "Discovers capabilities from the configured SearXNG instance via /config, including categories, engines, defaults, locales, and plugins.",
+    annotations: {
+        readOnlyHint: true,
+        openWorldHint: true,
+    },
+    inputSchema: {
+        type: "object",
+        properties: {
+            includeEngines: {
+                type: "boolean",
+                description: "Include enabled engine names in the response.",
+                default: false,
+            },
+            includeDisabled: {
+                type: "boolean",
+                description: "Include disabled engine names when includeEngines is true.",
+                default: false,
+            },
+            category: {
+                type: "string",
+                description: "Filter categories and engines to a single category name.",
+            },
+            refresh: {
+                type: "boolean",
+                description: "Bypass the process cache and fetch fresh /config data.",
+                default: false,
+            },
         },
+        required: [],
+    },
+};
+export const LITE_WEB_SEARCH_TOOL = {
+    name: "searxng_web_search",
+    description: "Web search. Returns titles, URLs, snippets.",
+    inputSchema: {
+        type: "object",
+        properties: { query: { type: "string", description: "Search query." } },
         required: ["query"],
     },
 };
+export const LITE_SUGGESTIONS_TOOL = {
+    name: "searxng_search_suggestions",
+    description: "Autocomplete search query suggestions.",
+    inputSchema: {
+        type: "object",
+        properties: { query: { type: "string", description: "Query prefix." } },
+        required: ["query"],
+    },
+};
+export const LITE_INSTANCE_INFO_TOOL = {
+    name: "searxng_instance_info",
+    description: "Discover SearXNG instance capabilities.",
+    inputSchema: {
+        type: "object",
+        properties: {},
+        required: [],
+    },
+};
+export const LITE_READ_URL_TOOL = {
+    name: "web_url_read",
+    description: "Fetch URL. Returns page text as markdown.",
+    inputSchema: {
+        type: "object",
+        properties: { url: { type: "string", description: "URL to fetch." } },
+        required: ["url"],
+    },
+};
 export const READ_URL_TOOL = {
     name: "web_url_read",
     description: "Fetches a URL and returns its text content converted to markdown. " +

package/dist/url-reader.d.ts

@@ -1,4 +1,5 @@
 import { McpServer } from "@modelcontextprotocol/sdk/server/mcp.js";
+import { type Dispatcher } from "undici";
 interface PaginationOptions {
     startChar?: number;
     maxLength?: number;
@@ -6,5 +7,7 @@ interface PaginationOptions {
     paragraphRange?: string;
     readHeadings?: boolean;
 }
+export declare const DEFAULT_MAX_CONTENT_LENGTH_BYTES: number;
+export declare function checkContentLength(mcpServer: McpServer, url: string, timeoutMs: number, dispatcher?: Dispatcher, baseRequestOptions?: RequestInit): Promise<number | null>;
 export declare function fetchAndConvertToMarkdown(mcpServer: McpServer, url: string, timeoutMs?: number, paginationOptions?: PaginationOptions): Promise<string>;
 export {};

package/dist/url-reader.js

@@ -8,6 +8,8 @@ import { getHttpSecurityConfig } from "./http-security.js";
 import { createURLFormatError, createURLSecurityPolicyError, createNetworkError, createServerError, createContentError, createConversionError, createTimeoutError, createEmptyContentWarning, createUnexpectedError } from "./error-handler.js";
 const REDIRECT_STATUS_CODES = new Set([301, 302, 303, 307, 308]);
 const MAX_REDIRECTS = 5;
+export const DEFAULT_MAX_CONTENT_LENGTH_BYTES = 5 * 1024 * 1024;
+const HEAD_TIMEOUT_CAP_MS = 3000;
 function isPrivateHostname(hostname) {
     const lower = hostname.toLowerCase().replace(/\.+$/, "");
     return lower === "localhost" || lower.endsWith(".localhost");
@@ -162,6 +164,53 @@ function applyPaginationOptions(markdownContent, options) {
     }
     return result;
 }
+export async function checkContentLength(mcpServer, url, timeoutMs, dispatcher, baseRequestOptions = {}) {
+    const controller = new AbortController();
+    const timeoutId = setTimeout(() => controller.abort(), Math.min(timeoutMs, HEAD_TIMEOUT_CAP_MS));
+    try {
+        const requestOptions = {
+            ...baseRequestOptions,
+            method: "HEAD",
+            signal: controller.signal,
+            redirect: "manual",
+        };
+        if (dispatcher) {
+            requestOptions.dispatcher = dispatcher;
+        }
+        const response = await undiciFetch(url, requestOptions);
+        const contentLength = response.headers.get("content-length");
+        if (!contentLength) {
+            return null;
+        }
+        const parsed = parseInt(contentLength, 10);
+        return Number.isNaN(parsed) || parsed < 0 ? null : parsed;
+    }
+    catch (error) {
+        logMessage(mcpServer, "warning", `HEAD check failed (proceeding with GET): ${error.message}`);
+        return null;
+    }
+    finally {
+        clearTimeout(timeoutId);
+    }
+}
+function getMaxContentLengthBytes(mcpServer) {
+    const rawValue = process.env.URL_READ_MAX_CONTENT_LENGTH_BYTES;
+    if (rawValue === undefined || rawValue.trim() === "") {
+        return DEFAULT_MAX_CONTENT_LENGTH_BYTES;
+    }
+    const parsed = parseInt(rawValue, 10);
+    if (Number.isNaN(parsed) || parsed <= 0) {
+        logMessage(mcpServer, "warning", `Ignoring invalid URL_READ_MAX_CONTENT_LENGTH_BYTES="${rawValue}". Expected a positive integer; using default ${DEFAULT_MAX_CONTENT_LENGTH_BYTES}.`);
+        return DEFAULT_MAX_CONTENT_LENGTH_BYTES;
+    }
+    return parsed;
+}
+function createContentTooLargeMessage(contentLength, maxBytes) {
+    const sizeMB = (contentLength / (1024 * 1024)).toFixed(1);
+    const limitMB = (maxBytes / (1024 * 1024)).toFixed(1);
+    return (`Content too large: server reports Content-Length of ${sizeMB} MB (limit: ${limitMB} MB). ` +
+        `Try using readHeadings or section to fetch only the relevant parts.`);
+}
 export async function fetchAndConvertToMarkdown(mcpServer, url, timeoutMs = 10000, paginationOptions = {}) {
     const startTime = Date.now();
     logMessage(mcpServer, "info", `Fetching URL: ${url}`);
@@ -184,6 +233,7 @@ export async function fetchAndConvertToMarkdown(mcpServer, url, timeoutMs = 1000
         throw createURLFormatError(url);
     }
     assertUrlAllowed(parsedUrl);
+    const maxContentLengthBytes = getMaxContentLengthBytes(mcpServer);
     // Create an AbortController instance
     const controller = new AbortController();
     const timeoutId = setTimeout(() => controller.abort(), timeoutMs);
@@ -216,6 +266,10 @@ export async function fetchAndConvertToMarkdown(mcpServer, url, timeoutMs = 1000
                 if (dispatcher) {
                     currentRequestOptions.dispatcher = dispatcher;
                 }
+                const contentLength = await checkContentLength(mcpServer, currentUrl.toString(), timeoutMs, dispatcher, currentRequestOptions);
+                if (contentLength !== null && contentLength > maxContentLengthBytes) {
+                    return createContentTooLargeMessage(contentLength, maxContentLengthBytes);
+                }
                 // Fetch the URL with the abort signal.
                 // Use undici's own fetch so it shares the same internal version as the
                 // Agent/ProxyAgent dispatcher — avoids the Node.js bundled-undici vs

package/dist/version.d.ts

@@ -1 +1 @@
-export declare const packageVersion = "1.3.4";
+export declare const packageVersion = "1.5.0";

package/dist/version.js

@@ -1 +1 @@
-export const packageVersion = "1.3.4";
+export const packageVersion = "1.5.0";

package/package.json

@@ -1,6 +1,6 @@
 {
     "name": "mcp-searxng",
-    "version": "1.3.4",
+    "version": "1.5.0",
     "mcpName": "io.github.ihor-sokoliuk/mcp-searxng",
     "description": "MCP server for SearXNG integration",
     "license": "MIT",