elevenlabs-voice-agent-mcp 1.0.0 → 1.0.1

package/.claude/settings.local.json

@@ -0,0 +1,23 @@
+{
+  "permissions": {
+    "allow": [
+      "WebSearch",
+      "Bash(git init:*)",
+      "Bash(git branch:*)",
+      "Bash(git add:*)",
+      "mcp__elevenlabs-voice-agents__elevenlabs_list_agents",
+      "mcp__elevenlabs-voice-agents__elevenlabs_list_phone_numbers",
+      "mcp__elevenlabs-voice-agents__elevenlabs_start_outbound_call",
+      "Bash(git checkout:*)",
+      "Bash(npx tsx:*)",
+      "Bash(npm run clean:*)",
+      "Bash(npm run build:*)",
+      "Bash(curl:*)"
+    ],
+    "deny": [],
+    "ask": []
+  },
+  "enabledMcpjsonServers": [
+    "elevenlabs-voice-agents"
+  ]
+}

package/.env.example

@@ -0,0 +1,3 @@
+# ElevenLabs API Configuration
+# Get your API key from: https://elevenlabs.io/app/settings/api-keys
+ELEVENLABS_API_KEY=your_elevenlabs_api_key_here

package/AGENTS.md

@@ -0,0 +1,37 @@
+# Repository Guidelines
+
+## Project Structure & Module Organization
+- `src/index.ts` boots the MCP server and registers tools; keep new exports centralized here.
+- `src/tools/*.ts` house tool implementations grouped by domain (agent, knowledge, outbound calling, batch jobs, phone numbers, utility).
+- `src/schemas/*.ts` contain Zod schemas for request/response validation; add or extend schemas before wiring new tools.
+- `src/services/elevenlabs-api.ts` wraps ElevenLabs HTTP calls; `src/services/formatters.ts` normalizes responses; `src/utils/*` holds shared helpers.
+- `dist/` stores compiled JS; edit TypeScript in `src/` and rebuild instead of touching `dist/`.
+
+## Build, Test, and Development Commands
+- `npm run dev` — run the server in watch mode via `tsx` for rapid iteration.
+- `npm run build` — type-check and emit JS to `dist/` using `tsc`.
+- `npm start` — run the compiled server (requires a fresh build).
+- `npm run clean` — remove `dist/` artifacts before fresh builds.
+- `npm test` — currently a stub; replace with real tests when added.
+- Set `ELEVENLABS_API_KEY` in `.env` (not committed) before running any command.
+
+## Coding Style & Naming Conventions
+- TypeScript ES modules, 2-space indentation, prefer explicit return types and narrow Zod schemas.
+- Tool identifiers follow the existing snake_case `elevenlabs_*` pattern to match client expectations; keep filenames descriptive (`outbound-tools.ts`, etc.).
+- Favor small, single-purpose helpers in `utils/` and reuse shared error handling patterns from `utils/error-handlers.ts`.
+- Keep public schemas and tool definitions colocated; avoid implicit `any` and keep imports relative within `src/`.
+
+## Testing Guidelines
+- No active test suite yet; when adding tests, keep them near sources (e.g., `src/tools/__tests__/`) and align names with the tool or schema under test.
+- Aim for request/response fixture coverage around new schemas and integration smoke tests for ElevenLabs API calls (use mocked HTTP).
+- Update `npm test` to run the chosen framework once added (Vitest/Jest are common fits for TS).
+
+## Commit & Pull Request Guidelines
+- Use concise, imperative commit subjects tied to scope (e.g., `Add batch call schema validation`, `Fix phone-number tool error handling`).
+- For PRs, include: summary of changes, commands run (`npm run build`, tests), any new env vars/config, and sample tool payloads/responses when behavior changes.
+- Avoid committing compiled `dist/` outputs or `.env`; prefer rebasing small topic branches for review clarity.
+
+## Security & Configuration Tips
+- Keep `ELEVENLABS_API_KEY` and any Twilio credentials in local `.env`; never log secrets.
+- When wiring into Claude Desktop, use absolute paths to `dist/index.js` and confirm Node 18+.
+- If adding new tools, document required permissions and expected side effects in the tool description so clients can surface them clearly.

package/CLAUDE.md

@@ -0,0 +1,233 @@
+# CLAUDE.md
+
+This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
+
+## Project Overview
+
+This is an MCP (Model Context Protocol) server that provides 23 specialized tools for developing and managing ElevenLabs Voice Agents. It enables Claude and other MCP clients to create, configure, test, monitor, and deploy AI-powered voice agents with outbound calling capabilities.
+
+## Development Commands
+
+### Build and Run
+```bash
+npm run build          # Compile TypeScript to JavaScript (outputs to dist/)
+npm run dev            # Development mode with auto-reload (uses tsx watch)
+npm start              # Run the compiled server from dist/
+npm run clean          # Remove build artifacts
+```
+
+### Environment Setup
+- Copy `.env.example` to `.env`
+- Set `ELEVENLABS_API_KEY` with your ElevenLabs API key
+- The server validates the API key on startup
+
+### Testing the Server
+After building, configure in Claude Desktop's config file (`~/Library/Application Support/Claude/claude_desktop_config.json` on macOS):
+```json
+{
+  "mcpServers": {
+    "elevenlabs-voice-agents": {
+      "command": "node",
+      "args": ["/absolute/path/to/dist/index.js"],
+      "env": {
+        "ELEVENLABS_API_KEY": "your_key_here"
+      }
+    }
+  }
+}
+```
+
+## Architecture
+
+### Core Structure
+
+The codebase follows a modular architecture with clear separation of concerns:
+
+1. **Entry Point** (`src/index.ts`)
+   - Initializes MCP server using `@modelcontextprotocol/sdk`
+   - Validates API key on startup
+   - Registers all tools from different modules
+   - Sets up stdio transport for communication
+
+2. **API Layer** (`src/services/elevenlabs-api.ts`)
+   - Central HTTP client for all ElevenLabs API requests
+   - Handles authentication via `xi-api-key` header
+   - Provides typed request helpers (GET, POST, PUT, PATCH, DELETE)
+   - 30-second request timeout
+
+3. **Tools Layer** (`src/tools/`)
+   - Each file exports MCP tool definitions with:
+     - `name`: Tool identifier
+     - `description`: What the tool does
+     - `inputSchema`: Zod schema for validation
+     - `handler`: Async function that executes the tool
+   - Tools are organized by functionality:
+     - `agent-tools.ts`: CRUD operations for voice agents
+     - `knowledge-tools.ts`: Adding knowledge bases to agents
+     - `tool-tools.ts`: Creating/managing webhook tools
+     - `conversation-tools.ts`: Retrieving conversation data
+     - `utility-tools.ts`: Voice listing, widget generation
+     - `outbound-tools.ts`: Single outbound call initiation
+     - `batch-calling-tools.ts`: Batch calling jobs management
+     - `phone-number-tools.ts`: Phone number import/management
+
+4. **Validation Layer** (`src/schemas/`)
+   - Zod schemas for runtime type validation
+   - Organized by domain (agents, tools, conversations, etc.)
+   - Defines input schemas used by MCP tools
+   - Reusable schema components in `common-schemas.ts`
+
+5. **Type System** (`src/types.ts`, `src/constants.ts`)
+   - TypeScript interfaces for all API entities
+   - Constants for supported models, languages, limits
+   - Type-safe enums derived from constant arrays
+
+### Data Flow
+
+1. MCP client (Claude) calls a tool with parameters
+2. Tool handler receives validated input (Zod schema)
+3. Handler calls API service with appropriate HTTP method
+4. API service authenticates and makes request to ElevenLabs
+5. Response is formatted (markdown or JSON) and returned
+6. Error handling catches and formats API errors
+
+### Key Design Patterns
+
+**Tool Structure**: Every tool follows this pattern:
+```typescript
+export const toolName = {
+  name: "elevenlabs_action_name",
+  description: "What it does",
+  inputSchema: zodSchema,
+  handler: async (args: SchemaType) => {
+    // 1. Make API request
+    // 2. Format response
+    // 3. Return formatted result
+  }
+};
+```
+
+**Response Formatting**: Tools support two formats:
+- `response_format: "markdown"` (default): Human-readable formatted output
+- `response_format: "json"`: Raw structured data
+
+**Pagination**: List endpoints use consistent pagination:
+- `limit`: Items per page (1-100, default 20)
+- `offset`: Skip N items
+- Response includes `has_more`, `next_offset`, `total`
+
+**Error Handling**: Centralized in `src/utils/error-handlers.ts`:
+- `handleElevenLabsError()`: Parses axios errors, provides actionable messages
+- `validateApiKey()`: Checks for API key on startup
+- HTTP status codes are mapped to user-friendly error messages
+
+**Response Truncation**: `src/utils/truncation.ts`:
+- Automatically truncates responses over 25,000 characters
+- Appends guidance on using pagination or filtering
+
+## Important Implementation Notes
+
+### API Authentication
+- All requests require `xi-api-key` header with ElevenLabs API key
+- Key is read from `process.env.ELEVENLABS_API_KEY`
+- Server exits with error if key is not set
+
+### Type Safety
+- Use strict TypeScript (`strict: true` in tsconfig.json)
+- ES2022 module system with Node16 resolution
+- All `.ts` imports must include `.js` extension (ESM requirement)
+
+### Tool Registration
+When adding new tools:
+1. Create tool definition in appropriate `src/tools/*.ts` file
+2. Create/update Zod schema in `src/schemas/*.ts`
+3. Add types to `src/types.ts` if needed
+4. Import and add to tools array in `src/index.ts`
+5. Rebuild with `npm run build`
+
+### Constants and Defaults
+- Default LLM: `gpt-4o-mini`
+- Default voice model: `eleven_flash_v2_5`
+- Default language: `en`
+- Character limit: 25,000
+- Request timeout: 30 seconds
+- Max batch recipients: 10,000
+
+### Outbound Calling Architecture
+- Requires Twilio integration
+- Phone numbers must be imported before use
+- Phone numbers must be assigned to agents
+- Batch calling supports up to 10,000 recipients
+- Supports dynamic variables via `conversation_initiation_client_data`
+  - Dynamic variables must be nested: `{dynamic_variables: {name: 'John', user_id: '123'}}`
+  - Can also include `conversation_config_override` for per-call customization
+  - Example: `{dynamic_variables: {customer_name: "Alice"}, conversation_config_override: {agent: {first_message: "Hi!"}}}`
+
+## Common Patterns
+
+### Making API Requests
+```typescript
+import { postRequest, getRequest } from "../services/elevenlabs-api.js";
+
+// POST request
+const result = await postRequest<ResponseType>(
+  "/convai/agents",
+  requestBody
+);
+
+// GET request with params
+const result = await getRequest<ResponseType>(
+  "/convai/agents",
+  { limit: 20, offset: 0 }
+);
+```
+
+### Creating Zod Schemas
+```typescript
+import { z } from "zod";
+
+export const MyToolInputSchema = z.object({
+  required_field: z.string(),
+  optional_field: z.string().optional(),
+  enum_field: z.enum(["option1", "option2"]),
+  response_format: z.enum(["markdown", "json"]).default("markdown")
+});
+
+export type MyToolInput = z.infer<typeof MyToolInputSchema>;
+```
+
+### Formatting Responses
+```typescript
+if (args.response_format === "json") {
+  return JSON.stringify(data, null, 2);
+}
+
+// Return markdown formatted string
+return `# Heading\n\n**Field**: ${data.field}\n`;
+```
+
+## Dependencies
+
+### Runtime
+- `@modelcontextprotocol/sdk`: ^1.6.1 - MCP server implementation
+- `axios`: ^1.7.9 - HTTP client for API requests
+- `zod`: ^3.23.8 - Runtime type validation
+
+### Development
+- `typescript`: ^5.7.2 - TypeScript compiler
+- `tsx`: ^4.19.2 - TypeScript execution and watch mode
+- `@types/node`: ^22.10.0 - Node.js type definitions
+
+### Requirements
+- Node.js 18 or higher (specified in package.json engines)
+
+## API Integration Notes
+
+- Base URL: `https://api.elevenlabs.io/v1`
+- All endpoints are under `/convai/` prefix
+- Rate limiting: 429 errors suggest waiting 60 seconds
+- Authentication: Header-based with `xi-api-key`
+- Conversation IDs start with `conv_`
+- Agent IDs start with `ag_`
+- Phone number IDs start with `pn_`
+- Batch IDs start with `batch_`

package/README.md

@@ -42,17 +42,10 @@ A Model Context Protocol (MCP) server for developing and managing ElevenLabs Voi
 - Node.js 18 or higher
 - ElevenLabs API key ([Get one here](https://elevenlabs.io))
 
-### Option 1: Install from npm (Recommended)
+### Setup
 
-```bash
-npm install -g elevenlabs-voice-agent-mcp
-```
-
-### Option 2: Install from source
-
-1. **Clone the repository:**
+1. **Clone or navigate to the project directory:**
    ```bash
-   git clone https://github.com/griffinwork40/elevenlabs-voice-agent-mcp.git
    cd elevenlabs-voice-agent-mcp
    ```
 
@@ -61,7 +54,13 @@ npm install -g elevenlabs-voice-agent-mcp
    npm install
    ```
 
-3. **Build the project:**
+3. **Set up environment variables:**
+   ```bash
+   cp .env.example .env
+   # Edit .env and add your ElevenLabs API key
+   ```
+
+4. **Build the project:**
    ```bash
    npm run build
    ```
@@ -85,26 +84,6 @@ Add this to your Claude Desktop configuration file:
 **macOS**: `~/Library/Application Support/Claude/claude_desktop_config.json`
 **Windows**: `%APPDATA%\Claude\claude_desktop_config.json`
 
-#### If installed via npm:
-
-```json
-{
-  "mcpServers": {
-    "elevenlabs-voice-agents": {
-      "command": "npx",
-      "args": [
-        "elevenlabs-voice-agent-mcp"
-      ],
-      "env": {
-        "ELEVENLABS_API_KEY": "your_elevenlabs_api_key_here"
-      }
-    }
-  }
-}
-```
-
-#### If installed from source:
-
 ```json
 {
   "mcpServers": {

package/call-exact-format.ts

@@ -0,0 +1,66 @@
+/**
+ * Replicates the EXACT successful call format from documentation
+ * No conversation_config_override - just dynamic variables
+ */
+
+import axios from 'axios';
+
+const apiKey = process.env.ELEVENLABS_API_KEY;
+
+if (!apiKey) {
+  console.error('❌ Error: ELEVENLABS_API_KEY environment variable not set');
+  process.exit(1);
+}
+
+async function makeTestCall() {
+  try {
+    console.log('🔄 Initiating call with EXACT format from documentation...\n');
+
+    // EXACT format from SUCCESSFUL_TEST_CALL_DOCUMENTATION.md
+    const response = await axios.post(
+      'https://api.elevenlabs.io/v1/convai/twilio/outbound-call',
+      {
+        agent_id: 'agent_2601kahgbheaftnr7150xp0x8jbf',
+        agent_phone_number_id: 'phnum_4501kayyj6xtfak80g09czq6yzxm',
+        to_number: '+16184072273',
+        conversation_initiation_client_data: {
+          dynamic_variables: {
+            lead_name: 'Connor',
+            rep_name: 'Connor',
+            lead_email: 'connor@atlasdigitalusa.com',
+            their_timezone: 'Central Time'
+          }
+        }
+      },
+      {
+        headers: {
+          'xi-api-key': apiKey,
+          'Content-Type': 'application/json'
+        }
+      }
+    );
+
+    console.log('✅ Call initiated successfully!\n');
+    console.log('📞 Call Details:');
+    console.log('─────────────────────────────────────');
+    console.log(`Conversation ID: ${response.data.conversation_id}`);
+    console.log(`Twilio Call SID: ${response.data.callSid}`);
+    console.log('─────────────────────────────────────\n');
+    console.log('Full Response:');
+    console.log(JSON.stringify(response.data, null, 2));
+
+  } catch (error) {
+    if (axios.isAxiosError(error)) {
+      console.error('❌ API Error:', error.response?.status);
+      console.error('Message:', error.response?.data?.detail || error.message);
+      if (error.response?.data) {
+        console.error('Full error:', JSON.stringify(error.response.data, null, 2));
+      }
+    } else {
+      console.error('❌ Unexpected error:', error);
+    }
+    process.exit(1);
+  }
+}
+
+makeTestCall();

package/call-wait-null.ts

@@ -0,0 +1,71 @@
+/**
+ * Test call with first_message set to null (not empty string)
+ * This should make the agent wait for human to speak first
+ */
+
+import axios from 'axios';
+
+const apiKey = process.env.ELEVENLABS_API_KEY;
+
+if (!apiKey) {
+  console.error('❌ Error: ELEVENLABS_API_KEY environment variable not set');
+  process.exit(1);
+}
+
+async function makeTestCall() {
+  try {
+    console.log('🔄 Testing call with first_message: null\n');
+
+    const response = await axios.post(
+      'https://api.elevenlabs.io/v1/convai/twilio/outbound-call',
+      {
+        agent_id: 'agent_2601kahgbheaftnr7150xp0x8jbf',
+        agent_phone_number_id: 'phnum_4501kayyj6xtfak80g09czq6yzxm',
+        to_number: '+16184072273',
+        conversation_initiation_client_data: {
+          dynamic_variables: {
+            lead_name: 'Connor',
+            rep_name: 'Connor',
+            lead_email: 'connor@atlasdigitalusa.com',
+            their_timezone: 'Central Time'
+          },
+          conversation_config_override: {
+            agent: {
+              first_message: null  // Try null instead of empty string
+            }
+          }
+        }
+      },
+      {
+        headers: {
+          'xi-api-key': apiKey,
+          'Content-Type': 'application/json'
+        }
+      }
+    );
+
+    console.log('✅ Call initiated!\n');
+    console.log('📞 Call Details:');
+    console.log('─────────────────────────────────────');
+    console.log(`Conversation ID: ${response.data.conversation_id}`);
+    console.log(`Twilio Call SID: ${response.data.callSid}`);
+    console.log('─────────────────────────────────────\n');
+    console.log('⚠️  Agent should wait for human to speak first\n');
+    console.log('Full Response:');
+    console.log(JSON.stringify(response.data, null, 2));
+
+  } catch (error) {
+    if (axios.isAxiosError(error)) {
+      console.error('❌ API Error:', error.response?.status);
+      console.error('Message:', error.response?.data?.detail || error.message);
+      if (error.response?.data) {
+        console.error('Full error:', JSON.stringify(error.response.data, null, 2));
+      }
+    } else {
+      console.error('❌ Unexpected error:', error);
+    }
+    process.exit(1);
+  }
+}
+
+makeTestCall();

package/create-agent-simple.ts

@@ -0,0 +1,84 @@
+/**
+ * Creates a new ElevenLabs voice agent using direct API call
+ * Simplified version that follows the working outbound call pattern
+ */
+
+import axios from 'axios';
+import fs from 'fs';
+
+const apiKey = process.env.ELEVENLABS_API_KEY;
+
+if (!apiKey) {
+  console.error('❌ Error: ELEVENLABS_API_KEY environment variable not set');
+  process.exit(1);
+}
+
+async function createAgent() {
+  try {
+    console.log('📖 Reading system prompt...\n');
+
+    // Read the system prompt from file
+    const systemPrompt = fs.readFileSync('demo-agent-system-prompt.xml', 'utf-8');
+
+    console.log('🔄 Creating agent via API...\n');
+
+    const requestBody = {
+      name: 'Positive Lead Outreach Agent',
+      conversation_config: {
+        agent: {
+          prompt: {
+            prompt: systemPrompt,
+            llm: 'gpt-4o-mini'
+          },
+          language: 'en'
+          // NOTE: first_message is INTENTIONALLY OMITTED so agent waits for human
+        },
+        tts: {
+          voice_id: '21m00Tcm4TlvDq8ikWAM',
+          model_id: 'eleven_flash_v2_5'
+        }
+      }
+    };
+
+    console.log('Request body structure:');
+    console.log(JSON.stringify(requestBody, null, 2).substring(0, 500) + '...\n');
+
+    const response = await axios.post(
+      'https://api.elevenlabs.io/v1/convai/agents',
+      requestBody,
+      {
+        headers: {
+          'xi-api-key': apiKey,
+          'Content-Type': 'application/json'
+        }
+      }
+    );
+
+    console.log('✅ Agent created successfully!\n');
+    console.log('📋 Agent Details:');
+    console.log('─────────────────────────────────────');
+    console.log(`Agent ID: ${response.data.agent_id}`);
+    console.log(`Name: ${response.data.name}`);
+    console.log('─────────────────────────────────────\n');
+
+    // Save agent ID
+    fs.writeFileSync('new-agent-id.txt', response.data.agent_id);
+    console.log(`💾 Agent ID saved to: new-agent-id.txt\n`);
+    console.log('Full response:');
+    console.log(JSON.stringify(response.data, null, 2));
+
+  } catch (error) {
+    if (axios.isAxiosError(error)) {
+      console.error('❌ API Error:', error.response?.status);
+      console.error('Message:', error.response?.data?.detail || error.response?.data?.message || error.message);
+      if (error.response?.data) {
+        console.error('Full error:', JSON.stringify(error.response.data, null, 2));
+      }
+    } else {
+      console.error('❌ Unexpected error:', error);
+    }
+    process.exit(1);
+  }
+}
+
+createAgent();

package/create-new-agent.ts

@@ -0,0 +1,98 @@
+/**
+ * Creates a new ElevenLabs voice agent for calling positive email leads
+ * Key feature: Agent waits for human to speak first (no first_message)
+ */
+
+import axios from 'axios';
+import fs from 'fs';
+import path from 'path';
+import { fileURLToPath } from 'url';
+
+const apiKey = process.env.ELEVENLABS_API_KEY;
+
+if (!apiKey) {
+  console.error('❌ Error: ELEVENLABS_API_KEY environment variable not set');
+  process.exit(1);
+}
+
+const __filename = fileURLToPath(import.meta.url);
+const __dirname = path.dirname(__filename);
+
+async function createAgent() {
+  try {
+    console.log('📖 Reading system prompt from demo-agent-system-prompt.xml...\n');
+
+    // Read the system prompt from file
+    const promptPath = path.join(__dirname, 'demo-agent-system-prompt.xml');
+    const systemPrompt = fs.readFileSync(promptPath, 'utf-8');
+
+    console.log('🔄 Creating new agent: "Positive Lead Outreach Agent"\n');
+    console.log('Configuration:');
+    console.log('  - LLM: gpt-4o-mini');
+    console.log('  - Voice: Rachel (21m00Tcm4TlvDq8ikWAM)');
+    console.log('  - Voice Model: eleven_flash_v2_5');
+    console.log('  - Language: en');
+    console.log('  - First Message: NONE (waits for human)\n');
+
+    const response = await axios.post(
+      'https://api.elevenlabs.io/v1/convai/agents',
+      {
+        name: 'Positive Lead Outreach Agent',
+        conversation_config: {
+          agent: {
+            prompt: {
+              prompt: systemPrompt,
+              llm: 'gpt-4o-mini'
+            },
+            // first_message is INTENTIONALLY OMITTED - agent will wait for human
+            language: 'en'
+          },
+          tts: {
+            voice_id: '21m00Tcm4TlvDq8ikWAM',
+            model_id: 'eleven_flash_v2_5'
+          }
+        }
+      },
+      {
+        headers: {
+          'xi-api-key': apiKey,
+          'Content-Type': 'application/json'
+        }
+      }
+    );
+
+    console.log('✅ Agent created successfully!\n');
+    console.log('📋 Agent Details:');
+    console.log('─────────────────────────────────────');
+    console.log(`Agent ID: ${response.data.agent_id}`);
+    console.log(`Name: ${response.data.name}`);
+    console.log(`Created At: ${response.data.created_at}`);
+    console.log('─────────────────────────────────────\n');
+
+    console.log('🔍 Conversation Config:');
+    console.log(JSON.stringify(response.data.conversation_config, null, 2));
+    console.log('\n');
+
+    console.log('⚠️  IMPORTANT: Save this Agent ID for making test calls!');
+    console.log(`   Agent ID: ${response.data.agent_id}\n`);
+
+    // Save agent ID to a file for easy reference
+    const agentIdFile = path.join(__dirname, 'new-agent-id.txt');
+    fs.writeFileSync(agentIdFile, response.data.agent_id);
+    console.log(`💾 Agent ID saved to: ${agentIdFile}\n`);
+
+  } catch (error) {
+    if (axios.isAxiosError(error)) {
+      console.error('❌ API Error:', error.response?.status);
+      console.error('Message:', error.response?.data?.detail || error.message);
+      if (error.response?.data) {
+        console.error('Full error:', JSON.stringify(error.response.data, null, 2));
+      }
+    } else {
+      console.error('❌ Unexpected error:', error);
+    }
+    process.exit(1);
+  }
+}
+
+createAgent();