Try Speak Free Book a Demo Login

MCP Server

Copied to clipboard!

MCP Server & CLI

Connect Claude, Cursor, Windsurf & AI Assistants to Your Speak AI Workspace

83 tools, 5 resources, 3 prompts, and 26 CLI commands. Transcribe, analyze, search, and manage media at scale from any MCP-compatible AI assistant or your terminal. Free and open source.

Get Started Free
Book a 15-min call
View on GitHub
Free 7-day trial. No credit card required. MCP server is free and open source (MIT).

Or pay as you go from $1.50/hr. No subscription required. See pricing.

MCP Endpoint
https://mcp.speakai.co/mcp
83
MCP Tools
26
CLI Commands
5
Resources
3
Prompts
70+
Languages

Installation

Connect your AI — choose your platform

Remote URL connection works with no installation. Or install locally via npm. The auto-setup wizard (speakai-mcp init) detects installed MCP clients and configures them automatically.

Claude (Web)
claude.ai browser


  1. Open claude.ai and go to Settings → Integrations.
  2. Click Add Integration and paste: https://mcp.speakai.co/mcp
  3. Authenticate with your Speak AI account and click Save. All 83 tools are now available in Claude.
Claude Desktop
Mac & Windows app


  1. Go to Settings → Developer → Edit Config and add:
    {
  "mcpServers": {
    "speakai": {
      "command": "npx",
      "args": ["-y", "@speakai/mcp-server"],
      "env": { "SPEAKAI_API_KEY": "YOUR_API_KEY" }
    }
  }
}
  2. Replace YOUR_API_KEY with your Speak AI API key.
  3. Save and restart Claude Desktop.
ChatGPT
OpenAI web & app


  1. Go to ChatGPT Settings → Integrations (Plus or Team required).
  2. Click Connect to MCP Server and enter: https://mcp.speakai.co/mcp
  3. Authenticate with your Speak AI credentials. Tools appear instantly.
Cursor
AI-first code editor


  1. Go to Cursor Settings → MCP and click Add MCP Server.
  2. Set the URL to: https://mcp.speakai.co/mcp
  3. Add your Speak AI API key and save. Tools appear in the AI panel.
VS Code
with GitHub Copilot


  1. Open VS Code Settings (JSON) and add:
    {
  "github.copilot.mcpServers": {
    "speakai": {
      "url": "https://mcp.speakai.co/mcp"
    }
  }
}
  2. Reload VS Code and open GitHub Copilot Chat. Speak AI tools appear automatically.
Claude Code CLI
Terminal · Official Plugin


  1. Recommended — install from the official marketplace. Type inside Claude Code:
    /plugin install speakai@claude-plugins-official
  2. Activate the plugin:
    /reload-plugins
  3. Run the getting-started skill to connect your Speak AI API key. Get your key from app.speakai.co/developers.
  4. Alternative — manual HTTP transport:
    claude mcp add speakai \
  --transport http \
  --url https://mcp.speakai.co/mcp
Any MCP Client (JSON)
Windsurf, Zed, Codeium…


  1. Find MCP / external tools settings in your client and add:
    {
  "servers": {
    "speakai": {
      "type": "http",
      "url": "https://mcp.speakai.co/mcp"
    }
  }
}
  2. Use HTTP/SSE transport and authenticate with your Speak AI API key.
  3. Not sure how? Email us — we’ll walk you through it.
npm (Local Install)
Auto-setup for all clients


  1. Install globally: npm install -g @speakai/mcp-server
  2. Run the auto-setup wizard: speakai-mcp init
  3. It detects Claude Desktop, Cursor, Windsurf, VS Code and configures them automatically.
  4. Set your API key: speakai-mcp config set-key

83 tools across 12 categories

Every Speak AI capability is available as an MCP tool. Your AI assistant can chain tools together for multi-step workflows: upload a recording, wait for processing, pull the transcript, extract action items, and export to PDF in a single conversation.

14 tools

Media

Upload from URL or local file, transcribe, get insights, export captions, update metadata, reanalyze with latest models. Upload-and-analyze in one call.

12 tools

Magic Prompt / AI Chat

Ask AI questions about any media, folder, or your entire workspace. Get chat history, manage favorites, export answers, and view usage statistics.

11 tools

Folders & Views

Create, clone, and organize folders. Save custom views with filters and sorting. Manage your media library structure programmatically.

10 tools

Recorder & Survey

Create embeddable recorders and surveys. Manage questions, branding, permissions. List submissions and generate shareable public URLs.

5 tools

Automations

Create and manage automation rules. Toggle automations on or off. Trigger workflows based on events in your workspace.

4 tools

Clips

Create highlight clips from time ranges across media files. Update, tag, and share clips. Pull the best moments from long recordings.

4 tools

Custom Fields

Define and manage custom metadata fields. Batch update fields across multiple media items for structured data workflows.

4 tools

Webhooks

Create webhook endpoints for real-time event notifications. Update, list, and manage webhooks for event-driven integrations.

4 tools

Meeting Assistant

Schedule the AI assistant to join meetings. List events, remove from active meetings, or cancel scheduled sessions. Works with Zoom, Google Meet, Teams.

4 tools

Media Embed

Create embeddable player widgets for your website. Update settings, check status, and get iframe URLs for any media file.

4 tools

Text Notes

Create text notes for AI analysis. Get NLP insights on any text. Re-run analysis with latest models. Update note content to trigger re-analysis.

5 tools

Exports, Search & Stats

Export as PDF, DOCX, SRT, VTT, TXT, CSV, or Markdown. Batch export with optional merge. Deep search across transcripts and insights. Workspace statistics and supported languages.

See all 83 tools in detail

26 CLI commands for scripting and automation

Everything the MCP server can do, the CLI can do from your terminal. Upload files, pull transcripts, search across your library, and pipe results to other tools. Every command supports --json for scripting.

Media management

upload, list-media, get-transcript, get-insights, status, export, update, delete, favorites, captions, reanalyze

AI and search

ask lets you query media, folders, or your whole workspace with AI. chat-history lists past conversations. search does full-text search across all transcripts and insights.

Organization and more

list-folders, create-folder, clips, clip, stats, languages, schedule-meeting, create-text. Plus config and init for setup.

# Upload a local file and wait for processing
speakai-mcp upload ./interview.mp3 -n “Q1 Interview” –wait

# Get plain-text transcript
speakai-mcp transcript abc123 –plain > meeting.txt

# Ask AI about a specific recording
speakai-mcp ask “What were the action items?” -m abc123

# Search all transcripts
speakai-mcp search “pricing concerns” –from 2026-01-01

# Export as PDF with speaker names
speakai-mcp export abc123 -f pdf –speakers

# List videos as JSON for piping
speakai-mcp ls –type video –json | jq ‘.mediaList[].name’

What you can do

Example workflows that combine multiple tools. Your AI assistant chains these automatically in a single conversation.

Transcribe and analyze a meeting

“Upload and transcribe this recording.” The assistant calls upload_and_analyze, waits for processing, and returns the full transcript with speaker labels, sentiment, action items, and key topics.

Research across your library

“What themes came up across all our customer interviews this month?” The assistant searches your media, retrieves insights from each file, and synthesizes patterns across 12+ recordings in one response.

Join a meeting and summarize

“Join my 2pm Zoom call and send me a summary with action items.” The assistant schedules the meeting bot, then pulls the transcript and insights after the meeting ends.

Build a weekly brief

“Prepare a brief from all meetings in the last week.” The assistant lists recent media, pulls insights from each, and creates a consolidated brief with decisions, action items grouped by owner, and follow-ups.

Create highlight clips

“Find the best quotes about pricing from the last 5 interviews and create clips.” The assistant searches transcripts, identifies time ranges, and creates shareable clips from each recording.

Batch export for a report

“Export all recordings from the Q1 Research folder as PDFs with speaker names.” The assistant lists folder contents and runs batch export, merging results into downloadable files.

How Speak AI compares

Most transcription tools offer 5–15 MCP tools focused on meeting notes. Speak AI provides a complete media intelligence platform with 83 tools, a full CLI, and capabilities no competitor offers.

CapabilitySpeak AITypical Meeting MCP
MCP tools835–15
CLI commands26None
MCP resources50–1
Built-in prompts30
NLP analytics (sentiment, keywords, entities)Yes, built inNo
Multi-model AI ChatClaude, Gemini, GPTSingle model or none
Cross-file analysisSearch and query across entire librarySingle meeting only
Upload from URL or local fileBothURL only or none
Export formatsPDF, DOCX, SRT, VTT, TXT, CSV, MarkdownText only
Embeddable recorder and surveysYes, 10 toolsNo
Automations and webhooksYes, 9 toolsNo
Meeting bot schedulingYes, 4 toolsLimited
Languages70+10–30
Open sourceMIT licenseVaries

Resources and built-in prompts

MCP resources provide direct data access without tool calls. Built-in prompts run multi-step workflows with a single command.

5 MCP Resources

Direct access to your media library, folders, supported languages, and per-file transcripts and insights. Clients can read these URIs without making explicit tool calls.

analyze-meeting prompt

Upload a recording and get a complete analysis: transcript, insights, action items, and key takeaways. One prompt runs the full pipeline.

research-across-media prompt

Search for themes, patterns, or topics across multiple recordings or your entire library. Specify a topic and optional folder to scope the research.

meeting-brief prompt

Prepare a brief from recent meetings. Pulls transcripts, extracts decisions, and summarizes open items. Defaults to the last 7 days.

Why Speak AI built the most comprehensive MCP server for media intelligence

MCP (Model Context Protocol) is the open standard that lets AI assistants like Claude and ChatGPT connect to external tools and data. Instead of copying and pasting transcripts into a chat window, you give your AI assistant direct access to your entire Speak AI workspace. It can upload files, pull transcripts, run NLP analysis, search across recordings, create clips, and export results, all through natural conversation.

Most transcription platforms that offer MCP servers provide a handful of tools focused on meeting notes. Speak AI takes a fundamentally different approach. With 83 MCP tools across 12 categories, plus 26 CLI commands, the Speak AI MCP server covers every capability of the platform: media management, AI-powered analysis, folder organization, embeddable recorders and surveys, automations, webhooks, meeting bot scheduling, custom fields, and multi-format exports.

The CLI changes how teams work with media data

The 26 CLI commands turn Speak AI into a scriptable media intelligence engine. Upload recordings from a shell script. Pull transcripts and pipe them to other tools. Search across your entire library from the command line. Every command supports --json output for integration with jq, Python scripts, or CI/CD pipelines. Teams use the CLI for batch processing, automated reporting, and building custom workflows that would be impractical through a web interface.

Built for researchers, not just meetings

Meeting transcription is table stakes. Speak AI is built for qualitative researchers, consulting firms, market research teams, and anyone who needs to analyze conversations at scale. The MCP server lets you query across dozens or hundreds of recordings: “What pricing concerns came up in customer interviews this quarter?” or “Compare sentiment across focus groups by region.” Cross-file analysis is the differentiator that single-meeting tools cannot match.

NLP analytics built into every tool

When you upload a recording through the MCP server, Speak AI does not just transcribe it. Every file gets automatic NLP analysis: sentiment scoring, keyword extraction, topic detection, theme identification, and named entity recognition. These structured insights are available as JSON through the get_media_insights tool. Your AI assistant can use them to compare sentiment across recordings, find trending topics, or pull specific data points without re-processing the audio.

Open source and extensible

The Speak AI MCP server is open source under the MIT license. View the full source on GitHub, install from npm, and extend it for your own workflows. The server authenticates using your personal API key and only accesses data in your Speak AI workspace. All communication uses HTTPS encryption. Full API documentation is available at docs.speakai.co.

Frequently asked questions

What is an MCP server?

MCP (Model Context Protocol) is an open standard that lets AI assistants connect to external tools and data. The Speak AI MCP server gives AI assistants like Claude, ChatGPT, Cursor, Windsurf, and VS Code direct access to 83 transcription, analysis, and media management tools in your Speak AI workspace.

Which AI assistants work with the Speak AI MCP server?

Claude (web, Desktop, and Code), ChatGPT, Cursor, Windsurf, VS Code with MCP extensions, and any AI tool that supports the MCP protocol. The auto-setup wizard (speakai-mcp init) detects installed clients and configures them automatically.

Do I need to be a developer to use this?

No. The remote connector setup for Claude.ai and ChatGPT requires no coding or installation. Add the remote MCP URL in settings and authenticate with your API key. The npm install is for local setups and the CLI.

What is the CLI and how is it different from the MCP server?

The CLI provides 26 commands you run directly in your terminal: speakai-mcp upload, speakai-mcp ask, speakai-mcp search, and more. The MCP server provides 83 tools that AI assistants call during conversation. Both are included in the same npm package.

Is my data secure?

Yes. The server authenticates using your personal API key and only accesses data in your Speak AI workspace. All communication uses HTTPS encryption. The server is open source under the MIT license, so you can audit the code on GitHub.

How much does it cost?

The MCP server and CLI are free and open source. You need a Speak AI account to use them. API access is available on all paid plans, and you get full access during the free 7-day trial with no credit card required. See pricing for plan details.

What languages are supported?

Speak AI supports transcription in over 70 languages with automatic language detection. All languages work through both the MCP server and CLI. Speaker diarization, timestamps, and NLP analytics are available across all supported languages.

Can I use both the MCP server and CLI together?

Yes, and this is the recommended approach for power users. Use the MCP server when you want your AI assistant to orchestrate complex workflows through conversation. Use the CLI for quick tasks, batch operations, and scripting. Both share the same API key and access the same workspace data.

Start using Speak AI from your AI assistant or terminal

83 MCP tools, 26 CLI commands, and the most comprehensive media intelligence integration available for AI assistants. Get started in 2 minutes.

Get started free

Create an account, grab your API key, and connect your AI assistant. Full access during the 7-day trial. No credit card required.

Start Free
Login
GitHub
npm
Developers
Automated Transcription
Integrations
Pricing