loading…
Search for a command to run...
loading…
An MCP server that enables AI assistants to search and query your LaunchBox game library across all platforms including Steam, GOG, and emulators. It allows use
An MCP server that enables AI assistants to search and query your LaunchBox game library across all platforms including Steam, GOG, and emulators. It allows users to check ownership for bundles, view game metadata, find duplicates, and analyze playtime statistics through natural language.
An MCP server that lets AI assistants query your LaunchBox game library. Search your entire game collection across all platforms (Steam, GOG, Epic, emulators, etc.) directly from your AI assistant.
Works with Claude Desktop, Claude Code, or any MCP-compatible client.
Requires Node.js 20+.
You'll need LaunchBox installed with your game data.
Add the server to your MCP client's configuration. There are two ways to do this:
Option A: Run directly from GitHub (no install needed)
{
"mcpServers": {
"launchbox": {
"command": "npx",
"args": ["-y", "github:danjam/mcp-launchbox"],
"env": {
"LAUNCHBOX_PLATFORMS_PATH": "C:/LaunchBox/Data/Platforms"
}
}
}
}
Option B: Clone and run locally
git clone https://github.com/danjam/mcp-launchbox.git
cd mcp-launchbox
npm install
Then point your MCP client at the local build:
{
"mcpServers": {
"launchbox": {
"command": "node",
"args": ["/path/to/mcp-launchbox/dist/index.js"],
"env": {
"LAUNCHBOX_PLATFORMS_PATH": "C:/LaunchBox/Data/Platforms"
}
}
}
}
Replace /path/to/mcp-launchbox with the actual path where you cloned the project.
Search the game library by title. Uses fuzzy matching across title and series fields. Punctuation (dashes, colons, &/and) is normalised before matching by default.
Confidence scale: 1.0 = perfect match, ≥0.85 = very likely, ≥0.65 = probable, <0.65 = speculative. Confidence is penalised when a query token only matches as a substring of a title token (e.g. "Doom" matching "Doomblade"). Results include exactMatch: true when the normalised query equals the normalised title. Use get_game_details for full metadata.
Tool: search_games
| Parameter | Type | Required | Description |
|---|---|---|---|
query |
string | Yes | Game title or series |
platform |
string | No | Platform (e.g. 'Windows', 'Arcade') |
installed |
boolean | No | true = installed only, false = uninstalled only, omit = all |
favorite |
boolean | No | true = favorites only, false = non-favorites only, omit = all |
limit |
number | No | Max results (default 25, max 100) |
exact |
boolean | No | Exact title match only, no fuzzy search (default false) |
Check which games from a list are already in the library. Uses normalised title matching first, then fuzzy matching — only matches with confidence ≥0.85 are included to avoid false positives. When exact is true, only exact title matches are returned with no fuzzy or prefix fallback. When no match is found, nearMisses shows up to 5 close candidates for diagnostics. A nearMiss with shorterTitle: true (no confidence field) means a shorter version of the title exists in the library (e.g. "Behind the Frame" for query "Behind the Frame: The Finest Scenery") — search the shorter title to confirm ownership. Punctuation is normalised before matching by default.
Results do not include storefront or version info — use get_game_details for that.
Designed for bundle duplicate checking — pass all the titles in one call instead of searching one at a time. Omit platform to check across all platforms (recommended for bundles).
Tool: check_library
| Parameter | Type | Required | Description |
|---|---|---|---|
games |
string[] | Yes | Game titles to look up |
platform |
string | No | Platform (e.g. 'Windows', 'Arcade') |
exact |
boolean | No | Exact title match only, no fuzzy search (default false) |
Get full details for a specific game by its ID. Returns all metadata in camelCase: title, platform, developer, genres (array), series, ratings (communityStarRating and starRating, both 0–5), playTime ({seconds, hours}), installed status, and more. String fields are null when empty/missing.
When a game has alternate versions (e.g. multiple storefronts like Steam/GOG/Epic, ROM regions, or platform ports), a versions array is included. Each entry has version (storefront name or region identifier), installed (boolean), and optionally region (normalised geographic label for ROM variants). Note: source only reflects the import origin of the primary entry, not all owned storefronts — check versions for the full picture.
Tool: get_game_details
| Parameter | Type | Required | Description |
|---|---|---|---|
id |
string | Yes | Game ID from search results |
include_notes |
boolean | No | Include notes (default false) |
List and filter games in the library. Unlike search_games, this does not do fuzzy matching — it returns games matching the specified filters, sorted and paginated. Use for browsing by platform, installed status, or favorites, and for recency queries like "what did I add this week." Returns total (filtered count) and results (compact game objects without confidence scores, but with dateAdded and lastPlayedDate timestamps).
Tool: list_games
| Parameter | Type | Required | Description |
|---|---|---|---|
platform |
string | No | Platform (e.g. 'Windows', 'Arcade') |
installed |
boolean | No | true = installed only, false = uninstalled only, omit = all |
favorite |
boolean | No | true = favorites only, false = non-favorites only, omit = all |
status |
string | string[] | No | Progress status (OR logic for arrays) |
sort |
string | No | Sort order; default "title". dateAdded/lastPlayedDate: most recent first; playTime: most played first |
limit |
number | No | Max results (default 25) |
offset |
number | No | Results to skip (default 0) |
List all platforms in the library with their game counts. No parameters needed.
Tool: list_platforms
Find duplicate game entries grouped by title — includes cross-platform and same-platform duplicates.
Tool: find_duplicates
| Parameter | Type | Required | Description |
|---|---|---|---|
query |
string | No | Optional title filter (fuzzy match) |
limit |
number | No | Max duplicate groups to return (default 25) |
offset |
number | No | Groups to skip (default 0) |
Pick a random game from the library, optionally filtered. Use for "what should I play?" or "surprise me" queries. Returns a single game (same shape as list_games results) plus matchPool showing how many games matched the filters. Returns null game when no matches.
Tool: random_game
| Parameter | Type | Required | Description |
|---|---|---|---|
platform |
string | No | Platform (e.g. 'Windows', 'Arcade') |
installed |
boolean | No | true = installed only, false = uninstalled only, omit = all |
favorite |
boolean | No | true = favorites only, false = non-favorites only, omit = all |
status |
string | string[] | No | Progress status (OR logic for arrays) |
Get library summary statistics (total games, total platforms, top 10 platforms by game count, and status counts). statusCounts lists all distinct progress values with their counts, sorted descending — use it to discover what free-form status values exist in the library. No parameters needed.
Tool: get_stats
Reload all game data from disk. Use after adding or removing games in LaunchBox. Returns game/platform counts plus added and removed arrays showing what changed since the previous load (id, title, platform for each). On first load, the diff is omitted. If the set of progress/status values changes, a notifications/tools/list_changed notification is sent so clients can re-fetch updated tool schemas. No parameters needed.
Tool: reload_library
Выполни в терминале:
claude mcp add mcp-launchbox -- npx Web content fetching and conversion for efficient LLM usage.
Retrieval from AWS Knowledge Base using Bedrock Agent Runtime.
автор: modelcontextprotocolProvides auto-configuration for setting up an MCP server in Spring Boot applications.
A very streamlined mcp client that supports calling and monitoring stdio/sse/streamableHttp, and can also view request responses through the /logs page. It also
автор: xuzexin-hzНе уверен что выбрать?
Найди свой стек за 60 секунд
Автор?
Embed-бейдж для README
Похожее
Все в категории ai