mcPER
Provides access to the ePER v84 genuine parts catalogue for Alfa Romeo vehicles, enabling VIN-based vehicle specification lookup, specification-aware part fitment, supersession history tracing, and parts diagrams with relevant parts highlighted.
Provides access to the ePER v84 genuine parts catalogue for Fiat vehicles, enabling VIN-based vehicle specification lookup, specification-aware part fitment, supersession history tracing, and parts diagrams with relevant parts highlighted.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@mcPER What's the rear shock absorber part number for VIN ZFA19800004275255?"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
mcPER
日本語 | English
Fiat / Lancia / Alfa Romeo / Abarth の純正部品カタログである ePER の v84 に登録された内容を AI に調べてもらうための MCP サーバーです。Claude Desktop、Claude Code、Cursor、VS Code(GitHub Copilot)などの MCP に対応したアプリから使うことができます。
例えば、VIN(車台番号)を伝えて「この車のリアのショックアブソーバーの部品番号は?」と頼むことで、AI に次の作業を依頼できます。
VIN から、車の仕様(グレード、製造日、エンジン、オプション)を割り出す
その仕様に合う部品だけを選ぶ。年式や仕様ごとの細かい切り替えも、ePER と同じルールで判定する
部品番号が新しい番号に置き換わっている場合は、その変遷を辿る
部品図を表示し、該当する部品の番号を赤枠で囲んで場所を示す
ePER の画面を開いて図を探す手間がなくなります。
非公式のツールです。 mcPER は Stellantis(Fiat Chrysler Automobiles)とは関係がなく、承認や支援も受けていません。Fiat、Lancia、Alfa Romeo、Abarth、ePER は各社の商標です。このリポジトリには ePER のデータは含まれていません。データを生成するには、ePER v84 をインストールした Windows PC が必要です。
必要なもの
MCP に対応したアプリ(このページでは Claude Desktop を例に説明します)
Node.js 22.13 以上
データを準備するとき: ePER v84 をインストールした Windows PC
データを一度作ってしまえば、ePER は不要です。データのディレクトリをコピーすることで、ePER を入れていない PC や Mac でも使えます。
Related MCP server: Vincario MCP Server
準備
1. データを生成する
ePER のある Windows で一度だけ実施する必要があります。
このディレクトリでコマンドプロンプトなどを開き、次を実行します。
npm install
npm run build-data数分で data/pack ディレクトリが生成されます。ディレクトリの中身は部品カタログ、VIN データ、部品図で、約 5GB です。説明文は日本語と英語だけを入れています。
ePER を既定以外の場所にインストールしている場合は、EPER_DIR でインストール先を指定します。
2. Claude Desktop に登録する
Claude Desktop の設定ファイルに次を追加します。
Windows:
%APPDATA%\Claude\claude_desktop_config.jsonMac:
~/Library/Application Support/Claude/claude_desktop_config.json
{
"mcpServers": {
"mcper": {
"command": "node",
"args": ["--no-warnings", "C:\\path\\to\\mcPER\\src\\index.js"]
}
}
}args の2つ目には、このディレクトリの src\index.js の場所を書きます(Mac なら /path/to/mcPER/src/index.js の形)。Claude Desktop が node を見つけられない場合は、command に node のフルパス(例: C:\\Program Files\\nodejs\\node.exe、/opt/homebrew/bin/node)を書いてください。
保存したら、Claude Desktop を完全に終了してから起動し直します。ウィンドウを閉じるだけでは足りないので、タスクトレイのアイコンから「Quit」を選んでください。チャット欄のツール一覧に「mcper」が出ていれば準備完了です。
ほかのアプリに登録する
どのアプリでも、登録する中身は同じです。起動コマンドは node、引数は --no-warnings と src/index.js の場所の2つです。
Claude Code: 次のコマンドで登録できます。
claude mcp add mcper -- node --no-warnings /path/to/mcPER/src/index.jsCursor、VS Code など: 各アプリの MCP 設定ファイルに、上の Claude Desktop の例と同じ
commandとargsを書きこみます。
部品図を表示する機能は、アプリがツールの結果に入った画像を表示できる場合だけ使用できます。
ほかの PC や Mac などで使う
Node.js を入れる(Mac なら
brew install nodeなど)このディレクトリをコピーして
npm installを実行するdata/packも、外付けディスクなどでコピーする上と同じように、使うアプリに登録する
data/pack をこのディレクトリの外に置いた場合は、設定に "env": { "MCPER_DATA_DIR": "/path/to/pack" } のように場所を書き足してください。
使い方
登録したアプリで、普通に話しかけます。
「VIN ZFA19800004275255 の車の情報を教えて」
「この車のフロントブレーキパッドの部品番号と、その変遷を調べて。図で場所も見せて」
「バルケッタのリアサスペンションの部品図を見せて」
「部品番号 4449309 は何に置き換わった?」
初めてツールを使うときは、アプリが実行してよいかを確認されます。
知っておいてほしいこと
データは 2014 年時点のものです。 それ以降の部品番号の置き換えは含まれていません。
在庫や注文できるかどうかは分かりません。 2014 年当時の在庫情報は今は役に立たないので、あえて出していません。注文できるかは販売店で確かめてください。
ePER の日本語の部品名は短い表記です(例: ブレーキパッド → 「パッド・セット」)。見つからないときは、AI が英語などで探し直します。
価格は扱っていません。
困ったとき
ツール一覧に mcper が表示されない: 設定ファイルのパスと JSON の書き方を確認し、アプリを完全に終了してから起動し直してください。Claude Desktop の場合、エラーの内容は
%APPDATA%\Claude\logs\mcp-server-mcper.log(Mac は~/Library/Logs/Claude/)に表示されます「database not found」と出る:
npm run build-dataを実行していないか、data/packの場所が異なります動作を確かめたい:
npm testとnpm run check-eperを実行すると、ePER の表示結果と一致するかを検証できます
ライセンス
プログラムのコードは MIT License です。データの読み方と適合判定のロジックは openPER(MIT License、Copyright (c) 2023 Christopher Reynolds)から移植しているため、その著作権表示も LICENSE に含めています。
このライセンスはプログラムだけに適用されます。ePER のデータ(data/pack の中身を含む)には適用されず、その著作権は Fiat 側にあります。
mcPER (English)
日本語 | English
mcPER is an MCP server that lets an AI look up the contents of ePER v84, the genuine parts catalogue for Fiat / Lancia / Alfa Romeo / Abarth. It works with any MCP-compatible app, such as Claude Desktop, Claude Code, Cursor and VS Code (GitHub Copilot).
For example, give the AI a VIN and ask "What is the part number of the rear shock absorber for this car?". It can then:
Work out the vehicle's specification (version, build date, engine, options) from the VIN
Pick only the parts that fit that specification, applying the same rules as ePER, including cut-overs by build date and specification
Follow the supersession history when a part number has been replaced by a newer one
Show the parts drawing with the reference numbers of the relevant parts framed in red
No more opening ePER and hunting through drawings.
This is an unofficial tool. mcPER is not affiliated with, endorsed by or supported by Stellantis (Fiat Chrysler Automobiles). Fiat, Lancia, Alfa Romeo, Abarth and ePER are trademarks of their respective owners. This repository does not contain any ePER data. Generating the data requires a Windows PC with ePER v84 installed.
Requirements
An MCP-compatible app (this page uses Claude Desktop as the example)
Node.js 22.13 or later
For preparing the data only: a Windows PC with ePER v84 installed
Once the data has been generated, ePER is no longer needed. Copy the data directory and mcPER also works on PCs and Macs without ePER.
Setup
1. Generate the data
This needs to be done once, on the Windows PC with ePER.
Open a command prompt in this directory and run:
npm install
npm run build-dataAfter a few minutes the data/pack directory is created. It contains the parts catalogue, the VIN data and the parts drawings, about 5 GB in total. Descriptions are included in Japanese and English only.
If ePER is installed somewhere other than the default location, set EPER_DIR to the installation directory.
2. Register with Claude Desktop
Add the following to the Claude Desktop configuration file:
Windows:
%APPDATA%\Claude\claude_desktop_config.jsonMac:
~/Library/Application Support/Claude/claude_desktop_config.json
{
"mcpServers": {
"mcper": {
"command": "node",
"args": ["--no-warnings", "C:\\path\\to\\mcPER\\src\\index.js"]
}
}
}The second item in args is the path to src\index.js in this directory (on a Mac, something like /path/to/mcPER/src/index.js). If Claude Desktop cannot find node, put the full path to node in command (e.g. C:\\Program Files\\nodejs\\node.exe or /opt/homebrew/bin/node).
After saving, quit Claude Desktop completely and start it again. Closing the window is not enough: choose "Quit" from the tray icon. When "mcper" appears in the chat's tool list, you are ready.
Registering with other apps
The registration is the same in every app: the command is node, and the two arguments are --no-warnings and the path to src/index.js.
Claude Code: register with this command:
claude mcp add mcper -- node --no-warnings /path/to/mcPER/src/index.jsCursor, VS Code, etc.: write the same
commandandargsas in the Claude Desktop example into the app's MCP configuration file.
Drawings can only be shown if the app can display images included in tool results.
Using it on another PC or a Mac
Install Node.js (on a Mac, e.g.
brew install node)Copy this directory and run
npm installCopy
data/packas well, e.g. with an external driveRegister it with your app as described above
If you put data/pack outside this directory, add its location to the configuration, e.g. "env": { "MCPER_DATA_DIR": "/path/to/pack" }.
Usage
Just talk to the app you registered it with:
"Tell me about the car with VIN ZFA19800004275255"
"Find the part number of the front brake pads for this car and its supersession history. Show me where it is on the drawing."
"Show me the rear suspension drawing for the Barchetta"
"What replaced part number 4449309?"
The first time a tool is used, the app asks for permission to run it.
Good to know
The data is from 2014. Supersessions after that are not included.
Availability is unknown. The 2014 stock information is of no use today, so it is deliberately left out. Check with a dealer whether a part can be ordered.
Japanese part names in ePER are terse (e.g. brake pads → "パッド・セット"). If nothing is found, the AI retries in English or with other words.
Prices are not covered.
Troubleshooting
mcper does not appear in the tool list: check the path in the configuration file and the JSON syntax, then quit the app completely and start it again. For Claude Desktop, errors are logged to
%APPDATA%\Claude\logs\mcp-server-mcper.log(on a Mac,~/Library/Logs/Claude/)"database not found":
npm run build-datahas not been run, ordata/packis in a different locationChecking that it works: run
npm testandnpm run check-eperto verify the results against what ePER displays
License
The program code is under the MIT License. The data access and fitment logic is ported from openPER (MIT License, Copyright (c) 2023 Christopher Reynolds), so its copyright notice is included in LICENSE as well.
This license applies to the program only. It does not apply to ePER data (including the contents of data/pack), which is copyrighted by Fiat.
Available Tools
14 toolsdecode_vinB
Decode a VIN using ePER chassis data: model, MVS version, catalogue(s), build date, engine number, organization number, colours and factory options (CARATT).
Inputs: vin (required), lang, include_options.
get_assembly_imageB
Show the exploded view (cliché) of an assembly part, e.g. a brake caliper with its pads and seals (parts with has_components or inside_assembly). Component reference numbers can be framed in red.
Inputs: part_number (required), view, highlight_refs, highlight_parts.
get_drawing_imageA
Show the exploded-view drawing of a parts table (same keys as get_table_parts). The reference numbers of the given parts are framed in red so the user can see where they are.
Inputs: group (required), sgs_code (required), subgroup (required), catalogue (required), +3 more.
get_modificationA
Describe a modification code (from a MODIF field such as "D2105") and its cut-over points (build date, chassis, engine, organization number).
Inputs: code (required), catalogue (required), lang.
get_partB
Get details of a part number: description, family, unit, weight, counts of supersessions, and the components if it is an assembly.
Inputs: part_number (required), lang.
get_part_historyB
Follow the supersession (replacement) chain of a part number in both directions: which numbers replaced it (with dates, quantity ratios and interchangeability) and which older numbers it replaced. current_parts lists the end(s) of the chain.
Inputs: part_number (required), lang, max_depth.
get_table_partsB
Get the parts list of a table (drawing): reference number, part number, quantity, description, notes, option pattern and modifications. Pass vin to evaluate fitment for that vehicle.
Inputs: group (required), sgs_code (required), subgroup (required), catalogue (required), +6 more.
list_cataloguesA
List catalogues (model ranges). Filter by sub-make (F=Fiat, T=Fiat Commercial, L=Lancia, R=Alfa Romeo, C=Abarth) and/or text in the description.
Inputs: make, query.
list_groupsB
List the main groups of a catalogue (engine, brakes, body, ...).
Inputs: catalogue (required), lang.
list_languagesA
List languages available in the catalogue.
Inputs: no named parameters.
list_subgroupsC
List the subgroups of a group in a catalogue.
Inputs: group (required), catalogue (required), lang.
list_tablesB
List the parts tables (drawings) of a subgroup, with their applicability. Pass vin to see which tables fit the vehicle.
Inputs: group (required), subgroup (required), catalogue (required), mvs, +3 more.
search_partsA
Search a catalogue for parts by description words (all words must match; Japanese or English) or by part number prefix. ePER Japanese names are terse katakana (e.g. "パッド セット" for brake pads, "フィルタ" for filters); if nothing matches, retry with a shorter word or English ("PADS", "OIL FILTER"), or browse groups. With vin the catalogue is taken from the VIN and results are sorted by fitment.
Inputs: query (required), mvs, vin, lang, +4 more.
where_usedC
List the catalogue tables where a part number is used, and assemblies/kits that contain it.
Inputs: part_number (required), lang, limit, catalogue.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 14 tools
The tools mostly target clearly distinct operations: hierarchical browsing (groups, subgroups, tables), content retrieval (table parts, part details), search, VIN decoding, supersession history, where-used, and images. The only mild overlaps are get_part/get_part_history/where_used (related but different facets) and get_drawing_image/get_assembly_image (both render exploded views but for different entity types), which descriptions clarify well.
Almost all tools follow a predictable snake_case verb_noun pattern (list_groups, get_table_parts, search_parts, decode_vin), with a few acceptable noun-centric variants like get_part_history and get_drawing_image. The outlier is where_used, which is a query phrase rather than a verb_noun name, but it remains readable and clear.
At 14 tools, the set is well-scoped for a read-only parts catalogue: it covers browsing, search, VIN decoding, part relationships, and imagery without bloat. Each tool has a distinct role and no redundant near-duplicates pad the surface.
The surface covers the full read-only lifecycle: catalogue navigation, VIN-based fitment, part lookup, supersession history, where-used, modification codes, and exploded views. Minor gaps remain—there is no direct lookup by vehicle attributes (e.g., model/year) without a VIN, and no explicit operation to resolve a table reference number to a part outside get_table_parts—but these are workarounds rather than dead ends.
Maintenance
Related MCP Connectors
Carbon Voice MCP serves as a bridge that connects AI assistants like ChatGPT, Claude, and Cursor to a user's Carbon Voice account, turning voice messages and conversations into a private, on-demand knowledge base. It provides 28 specialized tools for comprehensive voice messaging management, including creating and sending messages, accessing conversation history with instant transcription, running AI actions (summarization, TLDR generation, meeting notes), and managing workspace collaboration through folders, contacts, and team communications.
Generate AI images, videos, music, SFX & speech in any AI assistant. Results appear inline in chat.
KYB for AI agents: GLEIF entity, parent chain, sanctions and EU VAT checks. Data lookup only.
Draft, find, update and sign contracts in fynk, the European contract management platform. Your AI assistant can search and summarize your documents, create contracts from templates or PDFs, fill in parties, signatories and dynamic fields, build templates, manage tags, teams and metadata, and send documents out for e-signature (SES, AES, QES). It acts with your own fynk permissions via OAuth. Requires an Essential, Business or Pro plan (14-day free trial) with MCP enabled by an account admin.
Related MCP Servers
- AlicenseAqualityBmaintenanceProvides AI agents with instant, structured access to electronic component datasheets, pinouts, and electrical specifications without requiring PDF uploads. It enables seamless part searching, design validation, and side-by-side component comparisons across major hardware providers.1495 npm12MIT

Vincario MCP Serverofficial
FlicenseNot gradedqualityDmaintenanceEnables AI assistants to decode VINs, check stolen vehicle databases, and retrieve market valuations through natural language.2-- AlicenseNot gradedqualityDmaintenanceProvides AI agents with authoritative OEM automotive repair data including part numbers, torque specs, fluid capacities, and service procedures for vehicles 1982-2013 across 80+ makes, sourced from factory service manuals.MIT
- AlicenseAqualityCmaintenanceEnables LLMs to decode Vehicle Identification Numbers and look up vehicle makes, models, years, and specifications offline using a curated NHTSA vPIC database.81MIT