spotify-agent

Spotify Agentic Controller

A powerful agent-powered controller for Spotify built using the Model Context Protocol (MCP).
This project allows you to use natural language to control Spotify — play songs, manage playlists, get track info, and more — all through an intelligent agent interface.

Features

This agent uses spotify-mcp-server under the hood and can perform the following actions:

• Start, pause, or skip playback
• Search for tracks, albums, artists, or playlists
• Get detailed info about a track, album, artist, or playlist
• Manage and view your Spotify queue
• Create, update, or manage your playlists

---

Configuration & Setup

1. Get Spotify API Keys

To connect your Spotify account:

1. Go to developer.spotify.com
2. Log in and navigate to the Dashboard
3. Create a new app
4. Set the Redirect URI as:
   http://127.0.0.1:8080/callback
   (You can use any port, but must use HTTP and localhost/127.0.0.1)

Keep your Client ID and Client Secret safe — you'll need them later.

---

2. Clone the Repository

Clone the spotify-mcp repo locally:

git clone https://github.com/varunneal/spotify-mcp.git
cd spotify-mcp

Make sure you have Node.js installed (>=18) since this repo uses npx.

3. Run the Spotify MCP Server

From inside the cloned repo directory:

npx @modelcontextprotocol/inspector uv --directory . run spotify-mcp

This will start the Spotify MCP server and print an authentication link in the terminal.

readme_md_continued = """

4. Authenticate Your Spotify Account

• Click the authentication link shown in the terminal after running the above command.
• It will redirect you to Spotify for login/authorization.
• Tokens will be saved locally and reused for future sessions.

---

How to Run the Spotify Agent

Once you've completed the authentication and setup steps:

1. Install Python Requirements

pip install -r requirements.txt

Run the Agent Interface

streamlit run spotify-agent-final.py

This will launch a chat-style web interface using Streamlit with memory where you can interact with the agent.

Example Agent Capabilities

Here are some things you can say to the agent:

• "Pause the music" → Pauses playback
• "Play my chill playlist" → Finds a playlist you created and plays it
• "Add Blinding Lights to Gym Mix" → Searches and adds the song
• "Create a playlist called Vibes" → Makes a new playlist
• "What’s playing right now?" → Gets the current track
• "Tell me about the album After Hours" → Fetches detailed info
• "Play Blinding Lights from my Workout playlist" → Queues playlist, adds song again, and skips to it

---

Agent Behavior Notes

• If a song is requested from a playlist, the agent:
  • Queues the entire playlist
  • Adds the requested song again
  • Skips to that song
  • But only if the song exists in that playlist — otherwise, it pauses and asks the user
• When user says "my playlists", the agent prioritizes playlists owned by the user (jawahar)
• Agents automatically resolve vague inputs by searching and confirming actions

---

Built With

• spotify-mcp-server
• Model Context Protocol
• Agno Framework
• Google Gemini API

---

Future Ideas

• Voice input for real-time control
• Smart playlist recommendations via LLMs

---

License

This project is open-sourced under the Apache-2.0 License.