MCP Server & Claude Plugin

PyLipidParse can run as an MCP (Model Context Protocol) server, letting Claude Code, Claude Desktop, and any MCP-compatible AI assistant convert lipid names to molecular structures directly in conversation — no Python code required.

Quick start

Option 1: One-line setup (no pre-install)

Add to your .mcp.json or Claude Code MCP config:

{
  "mcpServers": {
    "pylipidparse": {
      "command": "uvx",
      "args": ["--from", "pylipidparse[mcp]", "pylipidparse-mcp"]
    }
  }
}

uv must be installed. The first run downloads and caches pylipidparse[mcp] automatically (~30–60 s due to RDKit).

Option 2: Install then run

pip install "pylipidparse[mcp]"   # requires Python 3.10+
pylipidparse-mcp                  # starts the stdio MCP server

Then point Claude Code at the installed command:

{
  "mcpServers": {
    "pylipidparse": {
      "command": "pylipidparse-mcp"
    }
  }
}

Or register it with the Claude Code CLI directly:

claude mcp add pylipidparse -- uvx --from "pylipidparse[mcp]" pylipidparse-mcp

Claude Code plugin

The easiest way to use PyLipidParse in Claude Code is via the plugin, which wires up the MCP server automatically.

Install from the Claude Code plugins directory at claude.com/plugins — search for PyLipidParse.

Or load locally from this repository:

git clone https://github.com/MontgomeryBohde/PyLipidParse.git
claude --plugin-dir ./PyLipidParse/claude-code-plugin

Once the plugin is active, you can ask Claude naturally:

What is the SMILES for PC 16:0/18:1(9Z)?
Give me the InChIKey for oleic acid (FA 18:1(9Z)).
Convert these 10 ceramides to SMILES and InChIKey: ...
Export these lipids as an SDF file.

Available tools

Tool	Description
`lipid_to_smiles`	Single lipid → canonical SMILES
`lipid_to_inchi`	Single lipid → InChI string
`lipid_to_inchikey`	Single lipid → InChIKey (27-char hash)
`batch_convert_lipids`	Multiple lipids → SMILES / InChI / InChIKey in one call
`lipid_to_mol_file`	Single lipid → MDL MOL file content (2D coordinates)
`lipids_to_sdf`	Multiple lipids → SDF file content (2D coordinates + metadata)

Input format

Lipid names must use full structural notation with explicit chain positions. Species-level notation is rejected because it does not define a unique structure.

Accepted	Rejected
`PC 16:0/18:1(9Z)`	`PC 34:1`
`FA 18:2(9Z,12Z)`	`FA 18:2`
`Cer 18:1;O2/16:0`	`Cer d18:1/16:0`
`TAG 16:0/18:1(9Z)/18:2(9Z,12Z)`	`TG 52:2`
`Cholesterol`	—

Supported dialects

Pass dialect to any tool to switch notation standard:

Dialect	Standard
`LipidMaps` (default)	LIPID MAPS shorthand
`Goslin`	Goslin shorthand
`SwissLipids`	SwissLipids notation
`HMDB`	Human Metabolome Database

Example: batch conversion

Convert these lipids to SMILES and InChIKey:
PC 16:0/18:1(9Z), PE 18:0/20:4(5Z,8Z,11Z,14Z), Cer 18:1;O2/16:0

Claude will call batch_convert_lipids with output_formats=["smiles","inchikey"] and return a table of results.

Error responses

If a conversion fails, the tool returns a structured error object rather than raising an exception, so batch jobs can continue past individual failures:

{
  "lipid_name": "PC 34:1",
  "error": "insufficient_detail",
  "message": "...",
  "hint": "Use full structural notation such as 'PC 16:0/18:1(9Z)'."
}

Error type	Cause
`parse_error`	Name not recognized by the selected dialect
`insufficient_detail`	Species-level notation — chain positions not specified
`unsupported_class`	Lipid class not yet supported by PyLipidParse
`structure_error`	Parsed successfully but molecule assembly failed