← 返回 Skills 市场

DeepGram Speech platform

Name: DeepGram Speech platform
Author: nerkn

作者 nerkn · GitHub ↗ · v1.0.1

cross-platform ✓ 安全检测通过

2358

总下载

当前安装

版本数

在 OpenClaw 中安装

/install deepgram

功能描述

Command-line tool for fast, accurate speech-to-text transcription from local files, URLs, or live audio using Deepgram’s API with customizable options.

使用说明 (SKILL.md)

Deepgram CLI Skills Guide

Tool

@deepgram/cli — command-line interface for Deepgram speech-to-text.

Install

npm install -g @deepgram/cli

Auth

deepgram login

Uses your Deepgram API key (stored locally).

Core Skill: Speech → Text

Transcribe a Local Audio File

deepgram listen prerecorded audio.wav

Transcribe with Options

deepgram listen prerecorded audio.wav \
  --model nova-2 \
  --language en \
  --punctuate \
  --diarize

Core Skill: Read / Reach Content

From URL (remote audio)

deepgram listen prerecorded https://example.com/audio.mp3

From STDIN (pipes)

cat audio.wav | deepgram listen prerecorded -

From Microphone (live)

deepgram listen microphone

Stop with Ctrl+C. Congrats, you just dictated reality.

Output Handling

Save Transcript

deepgram listen prerecorded audio.wav > transcript.json

Plain Text Output

deepgram listen prerecorded audio.wav --format text

Useful Flags (Memorize These)

--model – nova-2, general, etc.
--language – en, tr, de, …
--punctuate – adds punctuation
--diarize – speaker separation
--format – json, text, srt, vtt

Typical Workflow

Reach content (file / URL / mic)
Run deepgram listen
Capture output (JSON or text)
Post-process (search, summarize, subtitle)

Skill Summary

CLI-based speech-to-text
Local, remote, and live audio
Scriptable, pipe-friendly
Fast, accurate, no UI nonsense

Deepgram CLI: because keyboards are overrated.

安全使用建议

This skill appears to be a straightforward guide for using the official Deepgram CLI. Before installing or using it: 1) verify you're comfortable installing a global npm package and that @deepgram/cli is the official package (check the npm publisher and repository); 2) understand that audio and transcripts will be sent to Deepgram's servers — avoid sending sensitive audio unless your Deepgram account and policy permit it; 3) create and scope a separate Deepgram API key if you want to limit exposure, and revoke it if no longer needed; 4) be aware microphone access is needed for live capture and that the CLI stores your API key locally via its login flow. If you need the agent to run this skill autonomously, consider whether you trust the agent's prompts to send audio to an external cloud service.

功能分析

Type: OpenClaw Skill Name: deepgram Version: 1.0.1 The skill bundle is benign. It provides instructions for using the legitimate Deepgram CLI tool for speech-to-text. While it involves installing a global npm package, making network requests to fetch audio from URLs, and accessing the microphone, these capabilities are directly aligned with the stated purpose of a speech-to-text service and are not leveraged in a malicious or suspicious manner. There are no prompt injection attempts or instructions for the agent to perform actions outside of the tool's intended use, such as data exfiltration or unauthorized access.

能力评估

✓ Purpose & Capability

The SKILL.md describes CLI-based speech-to-text using @deepgram/cli and examples match the stated purpose (local files, URLs, stdin, microphone). There are no unrelated requirements (no extraneous cloud credentials or unrelated binaries).

ℹ Instruction Scope

Instructions stay within the transcription domain (install CLI, login, run deepgram listen). They implicitly transmit audio/transcripts to Deepgram's service (expected for this purpose) and require microphone/device access for live capture — users should note that audio and transcripts leave the local machine to the Deepgram service.

ℹ Install Mechanism

The registry contains no install spec; the README tells the user to run npm install -g @deepgram/cli. This is a normal, common install route but does involve pulling code from the npm registry and installing a global binary — users should verify the package publisher and be comfortable with global npm installs.

✓ Credentials

The skill declares no environment variables or secrets. The CLI does require a Deepgram API key (obtained via deepgram login) — this is proportional and expected for a speech-to-text integration; no unrelated secrets are requested.

✓ Persistence & Privilege

always is false and the skill does not request system-wide configuration changes. disable-model-invocation is false (agent may call it autonomously), which is the platform default — not a red flag by itself.

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 deepgram
安装完成后，直接呼叫该 Skill 的名称或使用 /deepgram 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

v1.0.1

- Migrated documentation from JavaScript SDK usage to new Deepgram CLI tool. - Updated installation and authentication instructions for CLI workflow. - Provided new command examples for transcribing local files, URLs, microphone input, and STDIN. - Added guidance on output handling, transcript formats, and common CLI flags. - Revised workflow and skill summary to emphasize command-line usage and scripting capabilities.

v1.0.0

- Initial release of the official JavaScript/TypeScript SDK for Deepgram’s speech and language AI services. - Supports real-time and batch speech-to-text, text-to-speech, voice agents, and audio analysis. - Provides simple client setup with API key or access token for authentication. - Includes examples for transcribing remote URLs, local files, and live audio streams. - Offers utilities for parsing transcript results and extracting text, timestamps, and word-level detail. - Supports Deepgram models, smart formatting, and provides SDK helpers for captions and proxy/browser usage.

元数据

Slug deepgram

版本 1.0.1

许可证 —

累计安装 4

当前安装数 4

历史版本数 2

常见问题