@yawlabs/tailscale-mcp 0.21.0 → 0.21.2

package/README.md

@@ -1,5 +1,9 @@
 # @yawlabs/tailscale-mcp
 
+[![Add to Yaw MCP](https://yaw.sh/yaw-mcp-button.svg)](https://yaw.sh/mcp/install?name=Tailscale&command=npx&args=-y%2C%40yawlabs%2Ftailscale-mcp&description=Manage%20your%20Tailscale%20tailnet%20-%20devices%2C%20ACLs%2C%20DNS%2C%20keys&source=https%3A%2F%2Fgithub.com%2FYawLabs%2Ftailscale-mcp)
+
+One click adds this to your local Yaw MCP config so it's available in every Yaw Terminal session. Or install manually below.
+
 [![npm version](https://img.shields.io/npm/v/@yawlabs/tailscale-mcp)](https://www.npmjs.com/package/@yawlabs/tailscale-mcp)
 [![License: MIT](https://img.shields.io/badge/License-MIT-blue.svg)](https://opensource.org/licenses/MIT)
 [![GitHub stars](https://img.shields.io/github/stars/YawLabs/tailscale-mcp)](https://github.com/YawLabs/tailscale-mcp/stargazers)
@@ -9,10 +13,6 @@
 
 Built and maintained by [Yaw Labs](https://yaw.sh).
 
-[![Add to Yaw MCP](https://yaw.sh/yaw-mcp-button.svg)](https://yaw.sh/mcp/install?name=Tailscale&command=npx&args=-y%2C%40yawlabs%2Ftailscale-mcp&description=Manage%20your%20Tailscale%20tailnet%20-%20devices%2C%20ACLs%2C%20DNS%2C%20keys&source=https%3A%2F%2Fgithub.com%2FYawLabs%2Ftailscale-mcp)
-
-One click adds this to your local Yaw MCP config so it's available in every Yaw Terminal session. Or install manually below.
-
 ## What's the point if the API already exists?
 
 You could `curl` the Tailscale API. The point isn't replacing `curl` — it's letting an agent compose multi-endpoint workflows in one turn without writing a script:
@@ -730,13 +730,13 @@ This shows a read-only banner in the Tailscale Admin Console pointing to your re
 
 ## Running on oam.js (optional)
 
-[oam.js](https://oamjs.org) runs this server unmodified, and the `tailscale-mcp` command only ever uses the **latest oam release, currently 0.15.2**. Verified against oam 0.15.2: full MCP handshake, all 97 admin-API tools plus `tailscale_tool_groups`, all 4 resources, identical error responses, and a clean stdout protocol stream — from the shipped bundle *and* straight from the TypeScript source with no build step.
+[oam.js](https://oamjs.org) runs this server unmodified, and the `tailscale-mcp` command only ever uses the **latest oam release, currently 0.18.0**. Verified against the published oam 0.18.0: full MCP handshake, all 97 admin-API tools plus `tailscale_tool_groups`, all 4 resources, identical error responses, and a clean stdout protocol stream — from the shipped bundle *and* straight from the TypeScript source with no build step.
 
-**oam 0.15.2 is the minimum.** A floor matters here: releases before 0.9.0 ran `child_process.execFile` arguments through a shell, re-splitting them on whitespace and executing shell metacharacters inside an argument, and this server shells out to the `tailscale` binary across its local-CLI tools, so that was a reachable bug rather than a theoretical one.
+**oam 0.18.0 is the minimum.** A floor matters here: releases before 0.9.0 ran `child_process.execFile` arguments through a shell, re-splitting them on whitespace and executing shell metacharacters inside an argument, and this server shells out to the `tailscale` binary across its local-CLI tools, so that was a reachable bug rather than a theoretical one.
 
 How the `tailscale-mcp` command (`bin/tailscale-mcp.mjs`) picks a runtime:
 
-- **`TAILSCALE_MCP_RUNTIME=auto`** (the default) — if a client already launched it with `oam run` on oam 0.15.2 or newer, the server runs in that process. Otherwise it uses `OAM_BIN` when that is 0.15.2 or newer, else asks every oam binary it can find — `%LOCALAPPDATA%\oam\bin` then `~/.oam/bin` on Windows, `~/.oam/bin` elsewhere, then `PATH` — for its version and uses the newest at or above the floor (on a tie the installed copy wins). With none, it runs on Node. An oam host older than 0.15.2 never serves the server itself: it hands off to the newest usable oam, or to Node on `PATH`, or exits with an error when there is neither. Whenever it looks for an oam, stderr names an `OAM_BIN` that was passed over and why; the oam binaries it found and passed over are named, each with its reason, only when no usable oam turns up.
+- **`TAILSCALE_MCP_RUNTIME=auto`** (the default) — if a client already launched it with `oam run` on oam 0.18.0 or newer, the server runs in that process. Otherwise it uses `OAM_BIN` when that is 0.18.0 or newer, else asks every oam binary it can find — `%LOCALAPPDATA%\oam\bin` then `~/.oam/bin` on Windows, `~/.oam/bin` elsewhere, then `PATH` — for its version and uses the newest at or above the floor (on a tie the installed copy wins). With none, it runs on Node. An oam host older than 0.18.0 never serves the server itself: it hands off to the newest usable oam, or to Node on `PATH`, or exits with an error when there is neither. Whenever it looks for an oam, stderr names an `OAM_BIN` that was passed over and why; the oam binaries it found and passed over are named, each with its reason, only when no usable oam turns up.
 - **`TAILSCALE_MCP_RUNTIME=oam`** — the same, but exit with an error instead of falling back to Node.
 - **`TAILSCALE_MCP_RUNTIME=node`** — always Node: in-process under `npx`, handed off to Node on `PATH` when a client launches the command with `oam run`.
 
@@ -761,7 +761,7 @@ The sandbox is applied by the `tailscale-mcp` command, which spawns a fresh oam
 }
 ```
 
-**Measure startup on your own hardware.** An MCP client cold-starts this server once per session, so startup is the cost that actually gets paid, and on the machine this was measured on node won it — 437ms vs 1554ms for `oam run` over 10 warmed runs (an earlier 5-run round showed 326ms vs 427ms; the box was busy, so treat the magnitude as noisy and the direction as the finding). Those runs used an oam that predates the 0.15.2 floor and have not been repeated since, so do not read them as a current ranking.
+**Measure startup on your own hardware.** An MCP client cold-starts this server once per session, so startup is the cost that actually gets paid, and on the machine this was measured on node won it — 437ms vs 1554ms for `oam run` over 10 warmed runs (an earlier 5-run round showed 326ms vs 427ms; the box was busy, so treat the magnitude as noisy and the direction as the finding). Those runs used an oam that predates the 0.18.0 floor and have not been repeated since, so do not read them as a current ranking.
 
 The published `tailscale-mcp` command prefers the newest usable oam it finds (see above). Without oam that costs almost nothing: discovery is file-existence checks only, never a subprocess, and the fallback runs the server inside the Node process npm already started. With oam installed, though, the command boots Node, runs `--version` on every oam binary it found to pick the newest, and only then boots oam, so it is always slower than pointing your client at a runtime directly — the config above for oam, `node /path/to/tailscale-mcp/dist/index.js` for Node. `TAILSCALE_MCP_RUNTIME=node` skips oam entirely.
 
@@ -796,4 +796,4 @@ Found a vulnerability? See [SECURITY.md](SECURITY.md) — please use GitHub's pr
 
 MIT
 
-[![Follow @TokenLimitNews on X](https://img.shields.io/badge/follow-%40TokenLimitNews-000000?logo=x&logoColor=white)](https://x.com/TokenLimitNews)
+[![Follow @YawLabs on X](https://img.shields.io/badge/follow-%40YawLabs-000000?logo=x&logoColor=white)](https://x.com/YawLabs)

package/bin/tailscale-mcp.mjs

@@ -91,7 +91,7 @@
  * shipped bundle; keep it in step.
  *
  * MINIMUM OAM VERSION
- * The latest oam release, 0.15.2 -- bump OAM_MIN when oam ships a newer one.
+ * The latest oam release, 0.18.0 -- bump OAM_MIN when oam ships a newer one.
  * Only the current oam is used and verified; an older one is passed over for a
  * newer oam, or for Node. The floor is not cosmetic: before 0.9.0
  * `child_process.execFile` ran its arguments through a SHELL, `exec` accepted
@@ -124,7 +124,7 @@ import { delimiter, join } from "node:path";
 import { fileURLToPath } from "node:url";
 
 /** The latest oam release, and the oldest one used. See MINIMUM OAM VERSION above. */
-const OAM_MIN = [0, 15, 2];
+const OAM_MIN = [0, 18, 0];
 
 /**
  * The oldest Node this package supports, matching package.json `engines.node`.

package/dist/index.js

@@ -4148,13 +4148,14 @@ var require_fast_uri = __commonJS({
         if (!malformedIPLiteral) {
           malformedHost = canonicalizeHost(parsed, options, schemeHandler, isIP2);
         }
-        if (!schemeHandler || schemeHandler && !schemeHandler.skipNormalize) {
-          if (uri.indexOf("%") !== -1) {
-            if (parsed.host !== void 0 && !malformedIPLiteral) {
-              const host = isIP2 ? parsed.host : normalizePercentEncoding(parsed.host, true);
-              parsed.host = reescapeHostDelimiters(host, isIP2);
-            }
+        if (uri.indexOf("%") !== -1 && parsed.host !== void 0 && !malformedIPLiteral) {
+          let host = isIP2 ? parsed.host : normalizePercentEncoding(parsed.host, true);
+          if (!isIP2) {
+            host = normalizePercentEncoding(host.toLowerCase());
           }
+          parsed.host = reescapeHostDelimiters(host, isIP2);
+        }
+        if (!schemeHandler || schemeHandler && !schemeHandler.skipNormalize) {
           if (parsed.path) {
             parsed.path = normalizePathEncoding(parsed.path);
           }
@@ -28700,6 +28701,9 @@ var Protocol = class {
       this.setRequestHandler(GetTaskPayloadRequestSchema, async (request, extra) => {
         const handleTaskResult = async () => {
           const taskId = request.params.taskId;
+          if (!await this._taskStore.getTask(taskId, extra.sessionId)) {
+            throw new McpError(ErrorCode.InvalidParams, `Task not found: ${taskId}`);
+          }
           if (this._taskMessageQueue) {
             let queuedMessage;
             while (queuedMessage = await this._taskMessageQueue.dequeue(taskId, extra.sessionId)) {
@@ -28730,12 +28734,12 @@ var Protocol = class {
             throw new McpError(ErrorCode.InvalidParams, `Task not found: ${taskId}`);
           }
           if (!isTerminal(task.status)) {
-            await this._waitForTaskUpdate(taskId, extra.signal);
+            await this._waitForTaskUpdate(taskId, extra.signal, extra.sessionId);
             return await handleTaskResult();
           }
           if (isTerminal(task.status)) {
             const result = await this._taskStore.getTaskResult(taskId, extra.sessionId);
-            this._clearTaskQueue(taskId);
+            this._clearTaskQueue(taskId, extra.sessionId);
             return {
               ...result,
               _meta: {
@@ -28772,7 +28776,7 @@ var Protocol = class {
             throw new McpError(ErrorCode.InvalidParams, `Cannot cancel task in terminal status: ${task.status}`);
           }
           await this._taskStore.updateTaskStatus(request.params.taskId, "cancelled", "Client cancelled task execution.", extra.sessionId);
-          this._clearTaskQueue(request.params.taskId);
+          this._clearTaskQueue(request.params.taskId, extra.sessionId);
           const cancelledTask = await this._taskStore.getTask(request.params.taskId, extra.sessionId);
           if (!cancelledTask) {
             throw new McpError(ErrorCode.InvalidParams, `Task not found after cancellation: ${request.params.taskId}`);
@@ -28900,6 +28904,19 @@ var Protocol = class {
     const handler = this._requestHandlers.get(request.method) ?? this.fallbackRequestHandler;
     const capturedTransport = this._transport;
     const relatedTaskId = request.params?._meta?.[RELATED_TASK_META_KEY]?.taskId;
+    const sessionId = capturedTransport?.sessionId;
+    const store = this._taskStore;
+    let relatedTaskFound = true;
+    let relatedTaskLookup;
+    if (relatedTaskId && store && this._taskMessageQueue && sessionId !== void 0) {
+      relatedTaskFound = false;
+      relatedTaskLookup = (async () => {
+        if (!await store.getTask(relatedTaskId, sessionId)) {
+          throw new McpError(ErrorCode.InvalidParams, `Task not found: ${relatedTaskId}`);
+        }
+        relatedTaskFound = true;
+      })();
+    }
     if (handler === void 0) {
       const errorResponse = {
         jsonrpc: "2.0",
@@ -28909,7 +28926,10 @@ var Protocol = class {
           message: "Method not found"
         }
       };
-      if (relatedTaskId && this._taskMessageQueue) {
+      if (relatedTaskId && relatedTaskLookup) {
+        const queuedError = { type: "error", message: errorResponse, timestamp: Date.now() };
+        relatedTaskLookup.then(() => this._enqueueTaskMessage(relatedTaskId, queuedError, sessionId), () => capturedTransport?.send(errorResponse)).catch((error51) => this._onerror(new Error(`Failed to send an error response: ${error51}`)));
+      } else if (relatedTaskId && this._taskMessageQueue) {
         this._enqueueTaskMessage(relatedTaskId, {
           type: "error",
           message: errorResponse,
@@ -28960,7 +28980,10 @@ var Protocol = class {
       closeSSEStream: extra?.closeSSEStream,
       closeStandaloneSSEStream: extra?.closeStandaloneSSEStream
     };
-    Promise.resolve().then(() => {
+    (relatedTaskLookup ?? Promise.resolve()).then(() => {
+      if (relatedTaskLookup && abortController.signal.aborted) {
+        throw new McpError(ErrorCode.ConnectionClosed, "Request was cancelled");
+      }
       if (taskCreationParams) {
         this.assertTaskHandlerCapability(request.method);
       }
@@ -28995,7 +29018,7 @@ var Protocol = class {
           ...error51["data"] !== void 0 && { data: error51["data"] }
         }
       };
-      if (relatedTaskId && this._taskMessageQueue) {
+      if (relatedTaskId && this._taskMessageQueue && relatedTaskFound) {
         await this._enqueueTaskMessage(relatedTaskId, {
           type: "error",
           message: errorResponse,
@@ -29475,7 +29498,7 @@ var Protocol = class {
       throw new Error("Cannot enqueue task message: taskStore and taskMessageQueue are not configured");
     }
     const maxQueueSize = this._options?.maxTaskQueueSize;
-    await this._taskMessageQueue.enqueue(taskId, message, sessionId, maxQueueSize);
+    await this._taskMessageQueue.enqueue(taskId, message, sessionId ?? this._transport?.sessionId, maxQueueSize);
   }
   /**
    * Clears the message queue for a task and rejects any pending request resolvers.
@@ -29504,12 +29527,13 @@ var Protocol = class {
    * Uses polling to check for updates at the task's configured poll interval.
    * @param taskId The task ID to wait for
    * @param signal Abort signal to cancel the wait
+   * @param sessionId Session of the request that waits, passed to the task store
    * @returns Promise that resolves when an update occurs or rejects if aborted
    */
-  async _waitForTaskUpdate(taskId, signal) {
+  async _waitForTaskUpdate(taskId, signal, sessionId) {
     let interval = this._options?.defaultTaskPollInterval ?? 1e3;
     try {
-      const task = await this._taskStore?.getTask(taskId);
+      const task = await this._taskStore?.getTask(taskId, sessionId);
       if (task?.pollInterval) {
         interval = task.pollInterval;
       }
@@ -30388,6 +30412,42 @@ var ExperimentalMcpServerTasks = class {
 };
 
 // node_modules/@modelcontextprotocol/sdk/dist/esm/server/mcp.js
+function toolInputElementCount(value, max) {
+  let count = 0;
+  const stack = [value];
+  while (stack.length > 0) {
+    const node = stack.pop();
+    if (node === null || typeof node !== "object")
+      continue;
+    if (Array.isArray(node)) {
+      for (const child of node) {
+        if (++count > max)
+          return count;
+        if (child !== null && typeof child === "object")
+          stack.push(child);
+      }
+    } else {
+      for (const key in node) {
+        if (!Object.prototype.hasOwnProperty.call(node, key))
+          continue;
+        if (++count > max)
+          return count;
+        const child = node[key];
+        if (child !== null && typeof child === "object")
+          stack.push(child);
+      }
+    }
+  }
+  return count;
+}
+function resolveMaxToolInputElements(value) {
+  if (value === void 0 || value === Infinity)
+    return void 0;
+  if (typeof value !== "number" || Number.isNaN(value) || value < 1) {
+    throw new RangeError(`maxToolInputElements must be a number of at least 1, or Infinity, got ${String(value)}`);
+  }
+  return value;
+}
 var McpServer = class {
   constructor(serverInfo, options) {
     this._registeredResources = {};
@@ -30399,6 +30459,7 @@ var McpServer = class {
     this._resourceHandlersInitialized = false;
     this._promptHandlersInitialized = false;
     this.server = new Server(serverInfo, options);
+    this._maxToolInputElements = resolveMaxToolInputElements(options?.maxToolInputElements);
   }
   /**
    * Access experimental features.
@@ -30529,12 +30590,15 @@ var McpServer = class {
    * Validates tool input arguments against the tool's input schema.
    */
   async validateToolInput(tool, args, toolName) {
+    if (this._maxToolInputElements !== void 0 && toolInputElementCount(args, this._maxToolInputElements) > this._maxToolInputElements) {
+      throw new McpError(ErrorCode.InvalidParams, `Invalid arguments for tool ${toolName}: arguments contain more than the maximum of ${this._maxToolInputElements} elements`);
+    }
     if (!tool.inputSchema) {
       return void 0;
     }
     const inputObj = normalizeObjectSchema(tool.inputSchema);
     const schemaToParse = inputObj ?? tool.inputSchema;
-    const parseResult = await safeParseAsync2(schemaToParse, args);
+    const parseResult = await safeParseAsync2(schemaToParse, args ?? {});
     if (!parseResult.success) {
       const error51 = "error" in parseResult ? parseResult.error : "Unknown error";
       const errorMessage = getParseErrorMessage(error51);
@@ -30772,7 +30836,7 @@ var McpServer = class {
       }
       if (prompt.argsSchema) {
         const argsObj = normalizeObjectSchema(prompt.argsSchema);
-        const parseResult = await safeParseAsync2(argsObj, request.params.arguments);
+        const parseResult = await safeParseAsync2(argsObj, request.params.arguments ?? {});
         if (!parseResult.success) {
           const error51 = "error" in parseResult ? parseResult.error : "Unknown error";
           const errorMessage = getParseErrorMessage(error51);
@@ -35225,7 +35289,7 @@ Install a newer Node (https://nodejs.org/en/download), or point your MCP client'
   );
   process.exit(1);
 }
-var version2 = true ? "0.21.0" : resolveVersionFallback();
+var version2 = true ? "0.21.2" : resolveVersionFallback();
 var subcommand = process.argv[2];
 var USAGE = `Usage: tailscale-mcp [command]
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@yawlabs/tailscale-mcp",
-  "version": "0.21.0",
+  "version": "0.21.2",
   "mcpName": "io.github.YawLabs/tailscale-mcp",
   "description": "Tailscale MCP server: admin-API tools for devices, ACLs, DNS, auth keys, users, and audit logs, plus a CLI to validate and deploy ACLs.",
   "license": "MIT",
@@ -64,13 +64,13 @@
   "overrides": {
     "hono": "^4.13.5",
     "@hono/node-server": "^1.19.15",
-    "fast-uri": "^3.1.6",
-    "ip-address": "^10.3.1",
+    "fast-uri": "^3.1.8",
+    "ip-address": "^10.7.3",
     "qs": "^6.16.0"
   },
   "devDependencies": {
     "@biomejs/biome": "^2.4.12",
-    "@modelcontextprotocol/sdk": "^1.30.0",
+    "@modelcontextprotocol/sdk": "^1.32.1",
     "@types/node": "^26.0.0",
     "esbuild": "^0.28.1",
     "postject": "^1.0.0-alpha.6",