MCP Server
Connect the Lotsofsounds sound effects API to Claude, Cursor, ChatGPT, and any AI agent with our free, open Model Context Protocol (MCP) server. Search, preview, and download royalty-free sounds from your AI assistant.
The sound effects MCP server for AI assistants
Lotsofsounds has a hosted Model Context Protocol (MCP) server. AI assistants can search, inspect, and download thousands of royalty-free (CC0) sound effects. Use the connection that your client supports.
- API key endpoint:
https://api.lotsofsounds.com/mcp - OpenAI OAuth endpoint:
https://api.lotsofsounds.com/mcp/openai - OpenAI plugin directory endpoint:
https://api.lotsofsounds.com/mcp/openai/plugin-directory - Transport: Streamable HTTP (stateless)
- Auth: a Lotsofsounds API key or a connected Lotsofsounds account
- Free tier:
browse_samplesworks with no key, so try it before you subscribe
Free to connect, Pro to search the full catalog
Browsing free sample sounds needs no key. Searching and downloading the full catalog requires a Pro or Enterprise plan. API clients use a key from the API Keys dashboard. OpenAI plugin users connect their Lotsofsounds account with OAuth.
Tools exposed
| Tool | What it does | Auth |
|---|---|---|
browse_samples | Returns curated free CC0 sample sounds | None |
search_sounds | Searches the full catalog by keyword, tags, category, and duration | Pro / Enterprise |
get_sound | Full metadata for a single sound | Pro / Enterprise |
download_sound | Time-limited signed download URL (1 hour) | Pro / Enterprise |
get_profile | Connected account identity for the OpenAI host | Connected account |
Example prompts
Once the server is connected, ask for sounds in plain words. The assistant turns your request into short keyword searches with tag and duration filters.
| Prompt | What the assistant calls |
|---|---|
| "Find me a cinematic boom" | search_sounds with q: "boom", tags: ["cinematic"] |
| "Find five funny sounds for this video" | search_sounds with q: "funny", limit: 5 |
| "Find a short UI success sound" | search_sounds with q: "success", max_duration: 2 |
| "Download the second one" | download_sound with the id from the last result |
Search is keyword based
search_sounds matches each keyword against sound names, descriptions and tags. Short keywords such as boom or chime return more results than full sentences.
Add it to Claude
In Claude, open Settings → Connectors, add a custom connector, and use this URL:
https://api.lotsofsounds.com/mcpAdd a request header named x-api-key and enter your Lotsofsounds API key as its value.
For a local Claude Desktop configuration, you can also use mcp-remote. Add this configuration to claude_desktop_config.json, and then restart Claude:
{
"mcpServers": {
"lotsofsounds": {
"command": "npx",
"args": [
"-y",
"mcp-remote",
"https://api.lotsofsounds.com/mcp",
"--header",
"x-api-key:${LOS_API_KEY}"
],
"env": { "LOS_API_KEY": "los_your_key" }
}
}
}Ask Claude "find me a short dramatic cinematic hit". Claude will call search_sounds for you.
Add it to Cursor
In ~/.cursor/mcp.json (or Settings → MCP → Add new server):
{
"mcpServers": {
"lotsofsounds": {
"url": "https://api.lotsofsounds.com/mcp",
"headers": { "x-api-key": "los_your_key" }
}
}
}Windsurf, VS Code with GitHub Copilot MCP, and other Streamable HTTP clients use the same URL and header values. Their configuration file format can be different.
Add it to Claude Code
claude mcp add lotsofsounds --transport http https://api.lotsofsounds.com/mcp \
--header "x-api-key: los_your_key"Use it with the OpenAI Agents SDK
The hosted MCP server works as a tool source for GPT agents:
from agents import Agent, Runner
from agents.mcp import MCPServerStreamableHttp
sounds = MCPServerStreamableHttp(
params={
"url": "https://api.lotsofsounds.com/mcp",
"headers": {"x-api-key": "los_your_key"},
}
)
agent = Agent(
name="Sound assistant",
instructions="Help the user find royalty-free sound effects.",
mcp_servers=[sounds],
)
result = await Runner.run(agent, "Find a gentle notification chime under 2 seconds")
print(result.final_output)Use it with ChatGPT
ChatGPT uses OAuth. It does not send a custom x-api-key header.
- In ChatGPT, open Settings → Security and login → Developer mode and enable developer mode.
- Open Settings → Plugins and select the add button.
- Enter
https://api.lotsofsounds.com/mcp/openai. - Sign in to Lotsofsounds and approve the connection.
The search_sounds, get_sound, and download_sound tools use the plan on the connected account. The public directory plugin uses a separate endpoint with the same tools and account rules.
Add it to Grok Build
Grok Build can use the API key endpoint with a fixed request header:
grok mcp add --transport http lotsofsounds https://api.lotsofsounds.com/mcp \
--header "x-api-key: ${LOS_API_KEY}"Grok also reads project MCP configuration from .mcp.json.
Add it to Gemini CLI
Use the Gemini CLI command and add the same request header:
gemini mcp add --transport http lotsofsounds https://api.lotsofsounds.com/mcp \
--header "x-api-key: ${LOS_API_KEY}"Gemini CLI stores the server configuration in its settings file.
How the paywall works
The MCP server is a thin client over the same REST API. It never widens access
beyond a caller's plan. Without API-key or OAuth access, protected catalog tools
ask the user to authenticate or point to pricing. The free
browse_samples tool always works. This
keeps the server safe to open-source and list in public MCP directories while the
catalog stays behind the Pro tier.
Related
- Sound Effects API - REST quickstarts, endpoints, and machine-readable resources
- API Reference - authentication, filters, responses, and errors
- LLM & AI Agent Integration - function-calling without MCP
- Getting Started - get your API key
- Pricing - Pro and Enterprise plans
Ready to integrate?
Get your API key and start building with thousands of sound effects and music tracks.