oremus-web-search 0.1.3 → 0.1.5

package/README.md

@@ -8,6 +8,13 @@ An MCP server that exposes:
 
 This is designed to be run with `npx` as an MCP server (stdio transport).
 
+## Client setup (Codex / Claude / others)
+
+- Codex CLI: see “Use in Codex CLI” below.
+- Claude Code: see `web-search-mcp/CLAUDE.md:1` or copy `web-search-mcp/.mcp.json.example:1` to your project as `.mcp.json`.
+- Copilot instructions: see `web-search-mcp/.github/copilot-instructions.md:1`.
+- Gemini instructions: see `web-search-mcp/GEMINI.md:1`.
+
 ## Why this exists
 
 - SearXNG is great for finding URLs.
@@ -90,6 +97,13 @@ Input:
 - `include_comments` (boolean, optional)
 - `include_tables` (boolean, optional)
 - `use_proxy` (boolean, optional)
+- `max_chars` (number, optional): cap returned text fields
+- `start_char` (number, optional): paging offset used with `max_chars`
+- `plain_text_fallback` (boolean, optional): if text/plain + extraction is empty, return raw body as text
+- `rewrite_github_blob_to_raw` (boolean, optional): rewrite GitHub `.../blob/...` URLs to `raw.githubusercontent.com/...`
+- `fetch_timeout_seconds` (number, optional): per-attempt HTTP timeout
+- `max_fetch_bytes` (number, optional): cap download size (may truncate HTML)
+- `max_total_seconds` (number, optional): best-effort overall time budget
 
 Output:
 

package/dist/index.js

@@ -7,7 +7,7 @@ import { StreamableHttpMcpClient } from "./streamableHttpMcpClient.js";
 import { webSearch } from "./searxng.js";
 import { FETCH_AND_EXTRACT_TOOL, ROTATE_VPN_TOOL, WEB_SEARCH_TOOL, isFetchAndExtractArgs, isRotateVpnArgs, isWebSearchArgs, } from "./types.js";
 import { createConfigResource, createHelpResource } from "./resources.js";
-const packageVersion = "0.1.3";
+const packageVersion = "0.1.5";
 async function main() {
     const config = loadConfig(process.env);
     const trafilaturaClient = new StreamableHttpMcpClient({
@@ -40,6 +40,15 @@ async function main() {
                 include_comments: args.include_comments ?? false,
                 include_tables: args.include_tables ?? false,
                 use_proxy: args.use_proxy ?? true,
+                max_chars: args.max_chars,
+                start_char: args.start_char,
+                max_fetch_bytes: args.max_fetch_bytes,
+                fetch_timeout_seconds: args.fetch_timeout_seconds,
+                user_agent: args.user_agent,
+                accept_language: args.accept_language,
+                plain_text_fallback: args.plain_text_fallback,
+                rewrite_github_blob_to_raw: args.rewrite_github_blob_to_raw,
+                max_total_seconds: args.max_total_seconds,
             });
             return { content: result.content };
         }

package/dist/streamableHttpMcpClient.js

@@ -1,3 +1,13 @@
+class McpHttpError extends Error {
+    status;
+    body;
+    constructor(status, body) {
+        super(`MCP HTTP ${status}: ${body.slice(0, 500)}`);
+        this.name = "McpHttpError";
+        this.status = status;
+        this.body = body;
+    }
+}
 function contentTypeBase(value) {
     if (!value)
         return "";
@@ -44,6 +54,11 @@ export class StreamableHttpMcpClient {
         this.url = options.url;
         this.authorization = options.bearerToken ? `Bearer ${options.bearerToken}` : undefined;
     }
+    resetSession() {
+        this.sessionId = undefined;
+        this.protocolVersion = undefined;
+        this.initializing = undefined;
+    }
     headers(extra) {
         const headers = {
             Accept: "application/json, text/event-stream",
@@ -70,7 +85,7 @@ export class StreamableHttpMcpClient {
         const ct = contentTypeBase(resp.headers.get("content-type"));
         if (resp.status >= 400) {
             const body = await resp.text().catch(() => "");
-            throw new Error(`MCP HTTP ${resp.status}: ${body.slice(0, 500)}`);
+            throw new McpHttpError(resp.status, body);
         }
         // Notifications often return 202 with no body.
         if (!("id" in message)) {
@@ -110,9 +125,9 @@ export class StreamableHttpMcpClient {
                 id: initId,
                 method: "initialize",
                 params: {
-                    protocolVersion: "2025-11-25",
+                    protocolVersion: "2025-03-26",
                     capabilities: {},
-                    clientInfo: { name: "oremus-web-search", version: "0.1.3" },
+                    clientInfo: { name: "oremus-web-search", version: "0.1.5" },
                 },
             };
             const initResp = await this.post(initReq);
@@ -136,22 +151,36 @@ export class StreamableHttpMcpClient {
         return this.initializing;
     }
     async callTool(name, args) {
-        await this.ensureInitialized();
-        const id = this.nextId++;
-        const req = {
-            jsonrpc: "2.0",
-            id,
-            method: "tools/call",
-            params: {
-                name,
-                arguments: args,
-            },
+        const attemptCall = async () => {
+            await this.ensureInitialized();
+            const id = this.nextId++;
+            const req = {
+                jsonrpc: "2.0",
+                id,
+                method: "tools/call",
+                params: {
+                    name,
+                    arguments: args,
+                },
+            };
+            const resp = await this.post(req);
+            if (!resp)
+                throw new Error("tools/call returned no response");
+            if (resp.error)
+                throw new Error(`tools/call error: ${resp.error.message}`);
+            return resp.result;
         };
-        const resp = await this.post(req);
-        if (!resp)
-            throw new Error("tools/call returned no response");
-        if (resp.error)
-            throw new Error(`tools/call error: ${resp.error.message}`);
-        return resp.result;
+        try {
+            return await attemptCall();
+        }
+        catch (err) {
+            if (err instanceof McpHttpError &&
+                err.status === 400 &&
+                /no valid session id/i.test(err.body)) {
+                this.resetSession();
+                return await attemptCall();
+            }
+            throw err;
+        }
     }
 }

package/dist/types.js

@@ -67,6 +67,42 @@ export const FETCH_AND_EXTRACT_TOOL = {
                 description: "Whether the Trafilatura service should route via its proxy pool.",
                 default: true,
             },
+            max_chars: {
+                type: "number",
+                description: "Maximum number of characters to return for extracted text fields.",
+            },
+            start_char: {
+                type: "number",
+                description: "Starting character offset for extracted text fields (used with max_chars).",
+            },
+            max_fetch_bytes: {
+                type: "number",
+                description: "Maximum number of bytes to download before extraction (may truncate HTML).",
+            },
+            fetch_timeout_seconds: {
+                type: "number",
+                description: "HTTP fetch timeout in seconds (per attempt).",
+            },
+            user_agent: {
+                type: "string",
+                description: "Override User-Agent header for the upstream fetch.",
+            },
+            accept_language: {
+                type: "string",
+                description: "Optional Accept-Language header for the upstream fetch.",
+            },
+            plain_text_fallback: {
+                type: "boolean",
+                description: "If the upstream is text/plain and extraction is empty, return the raw body as text.",
+            },
+            rewrite_github_blob_to_raw: {
+                type: "boolean",
+                description: "If the URL is a GitHub blob page, rewrite to raw.githubusercontent.com before fetching.",
+            },
+            max_total_seconds: {
+                type: "number",
+                description: "Maximum total time budget in seconds for all attempts (best-effort).",
+            },
         },
         required: ["url"],
     },

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "oremus-web-search",
-  "version": "0.1.3",
+  "version": "0.1.5",
   "description": "MCP server that combines SearXNG web search with Trafilatura extraction",
   "license": "MIT",
   "type": "module",