Skip to main content

What is HappyScribe MCP

Connect HappyScribe to Claude, ChatGPT, Cursor or Claude Code. Then search, summarize, transcribe, export and share your transcripts straight from the chat.

Written by Simon

HappyScribe MCP lets AI assistants like Claude and ChatGPT securely access your HappyScribe account.

MCP (Model Context Protocol) is an open standard that allows AI tools to connect with external services in a consistent way. Once connected, your AI assistant can search transcripts, summarize meetings, extract action items, transcribe new files, export and share transcripts, and manage glossaries and summary templates, all directly from your conversation.

Claude (claude.ai)

Step 1: Set up the HappyScribe Connector on Claude

Sign in to Claude at claude.ai, the connector page only opens for a signed-in account. Then, follow this link to open the connector page and click "Connect".

Step 2: Authorize your HappyScribe account

Sign in using your existing HappyScribe account and approve access.

This allows your AI assistant to access your transcription data only when requested in prompts.

Step 3: Start asking questions about your transcripts

Once connected, try prompts like:

  • "Show me my recent transcriptions"

  • "Summarize my last 3 meetings"

  • "Search all my transcriptions for mentions of quarterly revenue"


ChatGPT

Step 1: Install the HappyScribe plugin in ChatGPT

Follow this link and click "Install plugin".

Step 2: Authorize your HappyScribe account

Click "Sign in with HappyScribe" and use your existing HappyScribe account to approve access.

This allows ChatGPT to access your transcription data only when requested in prompts.

Step 3: Start asking questions about your transcripts

Once connected, try prompts like:

  • "Show me my recent transcriptions"

  • "Summarize my last 3 meetings"

  • "Search all my transcriptions for mentions of quarterly revenue"


Claude Desktop

Step 1: Open Developer settings

In the Claude Desktop app, open Settings and navigate to the Developer tab.

Step 2: Click "Edit Config" to open the configuration file and add the HappyScribe MCP server:

{
"mcpServers": {
"happyscribe": {
"type": "http",
"url": "https://www.happyscribe.com/mcp"
}
}
}

Step 3: Save the file and restart Claude Desktop. The OAuth authorization will trigger on first use.


Claude Code (Terminal)

Run this command in your terminal:

claude mcp add happyscribe --transport http https://www.happyscribe.com/mcp

Claude Code will handle the OAuth authorization flow automatically on first use.


Cursor

Step 1: Open MCP Settings

Go to Settings, then Tools & MCPs. And then Click on "Add Custom MCP"

Step 2: Paste the following config in the configuration file

{
"mcpServers": {
"happyscribe": {
"transport": "http",
"url": "https://www.happyscribe.com/mcp"
}
}
}

And make sure you save it with Command+S.

Step 3: Wait for the connector to appear and hit connect

Step 4: Authorize your Happyscribe Account

Step 5: Verify it's connected and start asking questions about your transcripts


What you can do with HappyScribe and AI assistants

After connecting, you can manage and explore transcripts using simple prompts.

Search across all transcripts

Find mentions of topics, people, or keywords across your recordings.

Example prompts:

  • "Search all my transcriptions for mentions of quarterly revenue"

  • "Find every time pricing was mentioned in customer calls"

  • "Show transcripts where John Smith spoke"

Summarize meetings

Ask your AI assistant to generate quick summaries from your transcripts.

Example prompts:

  • "Summarize my last 3 meetings"

  • "What were the main decisions in yesterday's meeting?"

  • "Give me action items from the last sales call"

Extract quotes with timestamps

You can retrieve exact quotes along with timestamps and editor links.

Example prompts:

  • "Find quotes where the customer mentioned pricing concerns"

  • "Show the most important quotes from my latest interview"

Analyze insights across multiple transcripts

AI assistants can analyze patterns across many recordings.

Example prompt:

"What did customers say about pricing across all sales calls this month?"

Upload and transcribe new recordings

You can create new transcripts directly through your AI assistant. There are three ways to send a file:

  • Attach a file to your message

  • Paste a public link, like a YouTube, Vimeo, Dropbox or Google Drive link

  • Use the upload box: ask "Show me the HappyScribe upload box" and drop your files in

Your assistant can also tell you when the transcript is ready and show it in the chat with an audio player.

Example prompt:

"Transcribe this interview in Spanish and save it in my Research folder."

For the full walkthrough, see Transcribe files from Claude or ChatGPT.

Export and share transcripts

Export transcripts as Word, PDF, subtitle (SRT, VTT), spreadsheet and other formats, and get a download link in the chat. Share them with teammates by email with view or edit rights, or turn on a public link.

Example prompts:

  • "Export my last interview as a Word document with speaker names"

  • "Share yesterday's planning meeting with sarah@company.com, view only"

Fix several words or names in one go

Correct a misspelled name or term everywhere in a transcript with a single request.

Example prompt:

"In my last interview, replace 'Jon' with 'John' and 'Happy Scribe' with 'HappyScribe'."

Create and edit glossaries

Create glossaries so names, brands and jargon are spelled right in every transcript, and apply them when you transcribe.

Example prompt:

"Look through my last 10 customer calls and create a glossary with the product and company names that come up."

For prompts and plan limits, see Create and edit glossaries

Set up meeting summary templates

Create summary templates, choose which one a meeting or series uses, and regenerate a summary with a different template.

Example prompts:

  • "Make a BANT template (Budget, Authority, Need, Timeline) for my discovery calls"

  • "Use my 1:1 template for all my future meetings"

Explore people and companies mentioned

HappyScribe automatically builds a knowledge graph from your transcripts.

You can ask your AI assistant to explore:

  • People mentioned

  • Companies mentioned

  • Related conversations across files

Example prompt:

"What did John Smith talk about across all our calls?"

Manage meetings

Your AI assistant can retrieve meeting details such as:

  • Calendar events

  • Recording status

  • Troubleshooting information

Search HappyScribe help articles


Available tools

These are the tools your AI assistant can use once HappyScribe is connected. You don't need to name them: describe what you want, and your assistant picks the right tool. Some apps, like Claude, let you choose which tools need your approval before they run.

Find and read transcripts

  • list_transcriptions: List your transcripts, filtered by folder, date, person or company

  • search_transcriptions: Search for exact words or names across your transcripts

  • get_transcription: Read one transcript, its summary or its details

  • get_transcriptions: Read up to 5 transcripts at once

  • get_folder_hierarchy: Show your folder structure

  • verify_quotes: Check quotes against the transcript and return timestamps and links to the editor

  • list_read_files: Show which files you've read this month and how many reads you have left on your plan

Transcribe files

  • create_transcription: Transcribe an attached file or a public link, or make subtitles

  • show_upload_box: Show an upload box in the chat to send files from your device

  • get_upload_url: Used by Claude to send an attached file to HappyScribe

  • finalize_upload: Used by Claude to confirm an attached file was sent

  • get_transcription_status: Check whether a transcript is ready

  • unlock_transcription: Continue a file on hold by spending credits, with your agreement

  • retranscribe: Transcribe an existing file again, for example in another language

  • show_transcript: Show a transcript in the chat with an audio player

Edit transcripts

  • replace_text_in_transcript: Replace a word or phrase everywhere in a transcript

  • replace_text_batch: Make up to 50 replacements at once, saved as one version

  • rename_speakers: Rename speakers, for example Speaker 1 to Alice

  • reassign_speakers: Assign a part of the transcript to a different speaker

  • rename_transcription: Rename a transcript

Export and share

  • export_transcription: Export up to 20 transcripts as Word, PDF, subtitles, spreadsheets and more

  • get_export_status: Get the download link for an export that's still processing

  • share_transcription: Invite people by email or turn the public link on or off

Organize files

  • create_folder: Create a folder

  • rename_folder: Rename a folder

  • delete_folder: Delete an empty folder

  • move_transcriptions: Move up to 50 transcripts to another folder

  • delete_transcriptions: Move up to 50 transcripts to the trash, where you can restore them

Glossaries

  • list_glossaries: List your workspace's glossaries

  • get_glossary: Show a glossary and its terms

  • create_glossary: Create a glossary

  • update_glossary: Rename a glossary or add, edit and remove terms

Meetings and summary templates

list_calendar_events: List your upcoming and past meetings and whether they'll be recorded

  • get_meeting_diagnostics: Find out why a meeting was or wasn't recorded

  • list_summary_templates: List the summary templates you can use

  • get_summary_template: Show a summary template

  • create_summary_template: Create a summary template

  • update_summary_template: Edit a summary template you created

  • delete_summary_template: Delete a summary template

  • set_meeting_template: Choose the template for a meeting, a series or all your future meetings

  • regenerate_summary: Create the meeting summary again, optionally with a different template

Workspace and help

  • get_workspace: Show details of the connected workspace

  • search_helpdesk: Search the HappyScribe Help Center

  • get_helpdesk_article: Read a Help Center article

  • request_missing_capability: Send the HappyScribe team a short, general note about something you asked for that isn't possible yet


Example prompts you can try

Once connected, you can interact with your transcripts using prompts like:

  • "Show me my recent transcriptions"

  • "Search all my transcriptions for mentions of quarterly revenue"

  • "Summarize my last 3 meetings"

  • "Find all quotes where the customer mentioned pricing concerns"

  • "What did John Smith talk about across all our calls?"

  • "Show me all transcriptions from last week in the Sales folder"

  • "Find every meeting where we talked about customer experience and share them with sarah@company.com, view only"

  • "Make a template for my 1:1s: wins, blockers, feedback given, next steps"


Authentication and permissions

HappyScribe protects your data using secure authentication.

  • HappyScribe uses OAuth 2.0 to connect with AI assistants

  • You authorize access using your existing HappyScribe account

  • Transcripts are accessed only when you ask your AI assistant to retrieve them


Privacy and data

Your data remains protected when using AI assistants.

  • HappyScribe's Privacy Policy applies to all data accessed through AI assistants

  • Transcript data is only sent to the AI assistant when you request it

  • HappyScribe doesn’t store your conversations with your AI assistant. We only receive the requests your assistant sends to HappyScribe, like a search or an export, and keep usage data about them to run and improve the service.

For more details, see our Security documentation.


Troubleshooting

My AI assistant cannot access my transcripts

Make sure:

  • You completed the OAuth authorization flow

  • You are logged into the correct HappyScribe account

  • The MCP server URL is correct

https://www.happyscribe.com/mcp

The AI assistant says it cannot find any transcripts

Possible reasons:

  • Your account has no transcripts yet

  • The assistant does not have permission to access your data

  • The MCP connection needs to be reauthorized

Try reconnecting the MCP server and signing in again.

My AI assistant cannot transcribe a video URL

Make sure:

  • The URL is publicly accessible

  • The link is correctly formatted

Example:

https://www.youtube.com/watch?v=vthirS86ovA

My attached file or transcript isn't going through

In Claude, attachments need code execution turned on. If your file is on hold, your workspace may be out of minutes. See Transcribe files from Claude or ChatGPT for both fixes.


Pro tip

Try asking your AI assistant:

"Summarize the last 3 meetings in this series and highlight recurring action items."

This is a great way to quickly spot patterns, decisions, and repeated topics across meetings.


​

Did this answer your question?