MCP server for LiquidSoap 2.4.0 documentation and script assistance
LiquidSoap MCP Server
Version-accurate Liquidsoap scripting using the official 2.4.0 docs.
LLMs get Liquidsoap wrong a lot. Mixing 1.x docs, outdated blog posts, and half-remembered API examples.
This MCP server fixes that.
It gives your AI assistant (Claude, ChatGPT, Cursor, Windsurf, etc.) real Liquidsoap 2.4.0 documentation, examples, and API references, making it finally possible to:
- Understand
.liqscripts - Generate correct 2.4.0 code
- Fix errors and deprecated usage
- Explore operators, functions, transitions, and patterns
- Build web radio pipelines with confidence
No hallucinations. No version drift. No mystery errors.
Why this exists
Liquidsoap is incredibly powerful, but:
- Documentation varies heavily between versions
- 1.x and 2.x syntax differs in subtle ways
- LLMs blend outdated examples into their answers
- Even the official docs are spread across sections, pages, and changelogs
This MCP server gives your AI one job:
Stick to Liquidsoap 2.4.0 exactly.
It exposes a clean, structured API around the official docs so your assistant becomes a reliable Liquidsoap expert.
Features
Version-Pinned Documentation (2.4.0)
- Full language reference
- Core API functions & operators
- Protocols (Icecast, HLS, HTTP, SRT, etc.)
- Encoder/decoder options
- Runtime settings
Smart Search
- Search functions/operators by name or keyword
- Search through examples, patterns, and cookbook items
- Search 2.4.0 changelog & migration notes
Script Assistance
- Detect deprecated functions (e.g.
null(),insert_metadata) - Warn about 1.x syntax
- Highlight common design pitfalls
Example Library
Ready-to-use snippets:
- Crossfading
- Fallback chains
- Harbor live input
- HLS output
- Cron scheduling (
cron.add,cron.parse) - Metadata rewriting
- LUFS normalization
- Blank detection
- Multi-output pipelines
Fast & Local
- Docs cached and indexed for instant responses
- No web requests needed once running
Demo (example)
Ask:
- “Explain what this Liquidsoap script does.”
- “Add a fallback before the HLS output.”
- “Show the docs for
crossfade.” - “Fix deprecated functions in this script.”
- “Rewrite this using Liquidsoap 2.4.0-style
null.”
Your AI responds using only the pinned 2.4.0 docs.
Installation
Prerequisites
- Node.js ≥ 18
- Any MCP-compatible client
Install via npm (recommended)
npm install -g liquidsoap-mcp-server
Or run from source
git clone https://github.com/splinesreticulating/LiquidSoapMCP.git
cd LiquidSoapMCP
npm install
npm run build
Usage
Claude Desktop Example
macOS:
~/Library/Application Support/Claude/claude_desktop_config.json
Windows:
%APPDATA%/Claude/claude_desktop_config.json
{
"mcpServers": {
"liquidsoap": {
"command": "npx",
"args": ["-y", "liquidsoap-mcp-server"]
}
}
}
From source:
{
"mcpServers": {
"liquidsoap": {
"command": "node",
"args": ["/absolute/path/to/LiquidSoapMCP/build/index.js"]
}
}
}
Docker:
{
"mcpServers": {
"liquidsoap": {
"command": "docker",
"args": ["run", "-i", "--rm", "liquidsoap-mcp-server"]
}
}
}
Build the image:
docker build -t liquidsoap-mcp-server .
Supported Tools
Core Tools
get_versionlist_sectionsget_documentation(section)search_functions(query)get_changelogget_examples(topic)validate_script_syntax(script)
Documentation Sections
languagereferenceprotocolssettingsencoding_formatsffmpegquickstartcookbook
Roadmap
- [ ] Liquidsoap script graph visualization
- [ ] Integrate
liquidsoap --checkfor full type validation - [ ] Multi-version switching (2.2.x, 2.3.x, etc.)
- [ ] Fuzzy search & semantic function lookup
- [ ] “Explain this error log” tool
- [ ] Pattern/snippet library
Contributing
PRs welcome. Especially:
- More examples
- Better search
- Doc parsing improvements
- Additional validation rules
- MCP integration templates
Resources
- https://www.liquidsoap.info/doc-2.4.0/
- https://modelcontextprotocol.io/
- https://github.com/modelcontextprotocol/typescript-sdk
Acknowledgments
Built for the Liquidsoap community, and for the DJs, webradio operators, self-hosters, and audio nerds who want Liquidsoap to be easier, safer, and more fun with AI assistance.
Want to support the project?
Please star the repo!
It helps others discover it and tells me this niche was worth carving out.
