← 返回 Skills 市场

Elevenlabs Conversational

Name: Elevenlabs Conversational
Author: aiwithabidi

作者 aiwithabidi · GitHub ↗ · v1.0.0

cross-platform ⚠ suspicious

428

总下载

当前安装

版本数

在 OpenClaw 中安装

/install elevenlabs-conversational

功能描述

Full ElevenLabs platform integration — text-to-speech, voice cloning, and Conversational AI agent creation. Not just TTS — build interactive voice agents wit...

使用说明 (SKILL.md)

🗣️ ElevenLabs Conversational

Not just TTS — full Conversational AI. Voice synthesis, cloning, and conversational AI agent creation for OpenClaw agents.

Voice Synthesis vs Conversational AI

Feature	Voice Synthesis (TTS)	Conversational AI
What	Text → Speech	Full voice agent
Flow	One-way	Bidirectional
Use case	Narration, alerts	Phone agents, assistants
Latency	Batch OK	Real-time required

Existing ElevenLabs skills only do TTS. This skill covers the full platform including Conversational AI agents.

Requirements

Variable	Required	Description
`ELEVENLABS_API_KEY`	✅	ElevenLabs API key

Quick Start

# List available voices
python3 {baseDir}/scripts/elevenlabs_api.py voices

# Text to speech
python3 {baseDir}/scripts/elevenlabs_api.py tts "Hello world" --voice Rachel --output hello.mp3

# TTS with emotion control
python3 {baseDir}/scripts/elevenlabs_api.py tts "I'm so excited!" --voice Rachel --stability 0.3 --style 0.8

# Streaming TTS (lower latency)
python3 {baseDir}/scripts/elevenlabs_api.py tts-stream "Hello world" --voice Rachel --output hello.mp3

# List conversational AI agents
python3 {baseDir}/scripts/elevenlabs_api.py list-agents

# Create a conversational AI agent
python3 {baseDir}/scripts/elevenlabs_api.py create-agent --name "Support Bot" --voice Rachel --prompt "You are a helpful support agent."

# Get agent details
python3 {baseDir}/scripts/elevenlabs_api.py get-agent \x3Cagent_id>

# Voice cloning (instant)
python3 {baseDir}/scripts/elevenlabs_api.py clone-voice "My Voice" --files sample1.mp3 sample2.mp3

Commands

`voices`

List all available voices with ID, name, category, and language.

`tts \x3Ctext>`

Convert text to speech (non-streaming).

--voice NAME — voice name or ID (default: Rachel)
--output FILE — output file path (default: output.mp3)
--model ID — model (default: eleven_multilingual_v2)
--stability FLOAT — 0.0-1.0, lower = more expressive (default: 0.5)
--similarity FLOAT — 0.0-1.0, voice similarity boost (default: 0.75)
--style FLOAT — 0.0-1.0, style exaggeration (default: 0.0)

`tts-stream \x3Ctext>`

Streaming TTS — lower latency, outputs as chunks arrive.

Same options as tts

`list-agents`

List all Conversational AI agents.

`create-agent`

Create a new Conversational AI agent.

--name NAME — agent name
--voice NAME — voice to use
--prompt TEXT — system prompt for the agent
--first-message TEXT — greeting message
--language CODE — language code (default: en)

`get-agent \x3Cagent_id>`

Get details of a conversational AI agent.

`clone-voice \x3Cname>`

Create an instant voice clone.

--files FILE [FILE ...] — audio samples (minimum 1, recommended 3+)
--description TEXT — voice description

Integration Patterns

With Twilio (Phone)

Create a Conversational AI agent
Configure Twilio webhook to point to ElevenLabs
Incoming calls route to your AI agent

With Vapi

Create voice in ElevenLabs
Use voice ID in Vapi assistant config
Vapi handles orchestration, ElevenLabs handles voice

With LiveKit

Generate TTS audio via streaming API
Publish audio track to LiveKit room
Subscribe to participant audio for STT pipeline

Credits

Built by M. Abidi | agxntsix.ai YouTube | GitHub Part of the AgxntSix Skill Suite for OpenClaw agents.

📅 Need help setting up OpenClaw for your business? Book a free consultation

安全使用建议

This skill appears to do what it says: it calls the official ElevenLabs API and only needs your ELEVENLABS_API_KEY. Before installing or running it: 1) Only provide audio files for cloning that you have rights to—voice-cloning can have legal/ethical implications. 2) Be aware audio files you supply for clone-voice will be uploaded to the ElevenLabs service; treat your API key like a secret and don't share it with untrusted code. 3) The clone-voice multipart implementation is handcrafted in the script and may be brittle—review/test in a safe environment. 4) If you are unsure who published this skill, consider obtaining the ElevenLabs API key from a limited-scope account or rotate the key after testing. Overall, the package is internally coherent and does not request unrelated privileges.

功能分析

Type: OpenClaw Skill Name: elevenlabs-conversational Version: 1.0.0 The skill is suspicious due to potential arbitrary file write and read vulnerabilities in `scripts/elevenlabs_api.py`. The script directly uses user-provided arguments for output file paths (`--output` in `cmd_tts`, `cmd_tts_stream`) and input file paths (`--files` in `cmd_clone_voice`) without apparent sanitization or path validation. This could allow a malicious agent prompt or user input to write to arbitrary locations on the filesystem or read sensitive files, respectively. While these capabilities are part of the skill's stated purpose (saving audio, cloning voices from files), the lack of input validation makes it a significant vulnerability, not malicious intent.

能力评估

✓ Purpose & Capability

The name/description (ElevenLabs TTS, voice cloning, conversational agents) matches the included script and SKILL.md examples: listing voices, TTS (including streaming), agent CRUD, and voice cloning. Required credential (ELEVENLABS_API_KEY) is appropriate for ElevenLabs API access and nothing unrelated is requested.

✓ Instruction Scope

SKILL.md instructs the agent to run the bundled Python script with arguments. The script only reads the declared ELEVENLABS_API_KEY, accesses audio files supplied by the user (for cloning), writes audio output files, and talks to the ElevenLabs API (api.elevenlabs.io). It does not read arbitrary system files, other environment variables, or send data to unknown endpoints. The Twilio/LiveKit integration sections are advisory patterns (no hidden webhooks or other network operations are invoked by the script).

✓ Install Mechanism

There is no install spec or external download; this is instruction-only with a bundled Python script. No external packages are fetched or arbitrary archives extracted. Running the script requires a local Python runtime but does not install software or create system-wide artifacts.

✓ Credentials

Only ELEVENLABS_API_KEY is required and declared as the primary credential. That is proportionate to the described functionality (API calls to ElevenLabs). No additional SECRET/TOKEN/PASSWORD vars or unrelated credentials are requested.

✓ Persistence & Privilege

The skill does not request always:true, does not modify other skills or agent system config, and does not require persistent system-level privileges. It runs on demand and only performs API calls and local file reads/writes as described.

如何使用

确保已安装 OpenClaw（本地或 Docker 部署）
在对话框中输入安装命令：/install elevenlabs-conversational
安装完成后，直接呼叫该 Skill 的名称或使用 /elevenlabs-conversational 触发
根据 Skill 的参数说明提供必要输入，即可获得结构化输出

版本历史

v1.0.0

Initial release: Full ElevenLabs platform integration for OpenClaw agents. - Supports text-to-speech, streaming TTS, voice cloning, and creation of Conversational AI agents. - Enables emotion control, interactive voice agent building, and phone system integration. - Command-line tools provided for voice selection, cloning, agent creation, and management. - Requires Python 3.10+ and an ElevenLabs API key.

元数据

Slug elevenlabs-conversational

版本 1.0.0

许可证 —

累计安装 0

当前安装数 0

历史版本数 1

常见问题