// mcp server

coming soon

use shrp directly inside claude, cursor, windsurf, and any mcp-compatible ai tool — no api calls, no glue code. add shrp once to your config and your ai agent can transcribe audio, pull youtube transcripts, and generate speech natively.

// what is mcp

model context protocol

mcp is an open standard that lets ai tools like claude and cursor call external capabilities as native tools — no custom integration required.

tools, not endpoints

instead of writing fetch() calls, your ai agent calls shrp tools directly. youtube_transcript, text_to_speech, transcribe — as natural as built-in functions.

one config, everywhere

add your shrp api key to your mcp config once. every project that uses that config gets access to shrp automatically.

// tools

youtube_transcriptlive

Extract a transcript from any YouTube video URL. Returns plain text or SRT.

url: stringformat?: "text" | "srt"
text_to_speechlive

Convert text to speech using Google Neural voices. Returns base64 MP3 audio.

text: stringvoiceName: stringlanguageCode: stringspeed?: number
transcribecoming soon

Submit an audio or video file URL for AI transcription via AssemblyAI. Returns a job ID to poll for results.

fileUrl: stringlanguage?: stringspeakerLabels?: boolean

// setup

add shrp to your mcp config with your api key. the server url and auth header are the same regardless of which client you use.

mcp config (any client)
{
  "mcpServers": {
    "shrp": {
      "url": "https://shrp.app/api/mcp",
      "headers": {
        "Authorization": "Bearer shrp_live_..."
      }
    }
  }
}
Claude Desktop
claude_desktop_config.json
~/Library/Application Support/Claude/
Cursor
.cursor/mcp.json
project root or ~/.cursor/
Windsurf
.codeiumrc
project root
Any MCP client
mcp config
per-client docs

// how it works in practice

1
generate an api key

go to /dashboard/api-keys and generate a shrp_live_ key. free accounts get 2 keys.

2
add to your mcp config

paste the config snippet above into your claude desktop, cursor, or windsurf mcp settings with your key.

3
use it naturally

tell your ai agent "get me the transcript of this video" or "convert this to speech" — it calls shrp automatically.

// why mcp over the rest api

no glue code

the rest api requires you to write fetch calls and handle responses. mcp tools are called directly by the ai — zero boilerplate.

works mid-conversation

your ai agent can pull a youtube transcript or generate audio mid-task, without leaving the conversation or switching tools.

composable

chain tools together. transcribe audio → summarise → convert summary to speech. all in one agent flow.

same auth, same quota

uses your existing api key and plan. no new accounts, no separate billing.

// get notified

the mcp server is in development. when it ships, it will appear in the changelog. in the meantime, you can use the rest api — the same endpoints, the same auth, the same quota.

get api key →rest api docs