steam-mcp
Enables AI assistants to query Steam profiles, libraries, achievements, store data, and reviews via the Steam Web API.
README
steam-mcp
An MCP server that exposes the Steam Web API to any AI harness — Claude Desktop, Claude Code, Cursor, Cline, Windsurf, or anything that speaks the Model Context Protocol.
Ask your assistant things like "How many hours have I put into Team Fortress 2?", "What are the most-played games my friend owns?", or "Summarize the reviews for Baldur's Gate 3" — and it fetches the answer live from Steam.
Why
The Steam Web API is powerful but sprawling: dozens of interfaces, inconsistent shapes, 64-bit SteamIDs, an undocumented store API, and a mandatory key for anything player-related. steam-mcp wraps the useful parts behind a small, well-documented set of workflow-oriented MCP tools that return LLM-friendly, token-efficient responses.
Quickstart
You need a Steam Web API key for player/profile tools. Get one (free) at https://steamcommunity.com/dev/apikey. Store tools (app details, reviews, search) work without a key.
Run it straight from GitHub with no install using uv
(works today — PyPI release pending):
STEAM_API_KEY=xxxxxxxx uvx --from git+https://github.com/rachittshah/steam-mcp steam-mcp
Or install it from source:
uv pip install git+https://github.com/rachittshah/steam-mcp
# once published to PyPI: uv pip install steam-mcp
Claude Desktop / Claude Code
Add to your MCP config (claude_desktop_config.json, or .mcp.json for Claude Code):
{
"mcpServers": {
"steam": {
"command": "uvx",
"args": ["--from", "git+https://github.com/rachittshah/steam-mcp", "steam-mcp"],
"env": { "STEAM_API_KEY": "your-key-here" }
}
}
}
(After a PyPI release the args simplify to ["steam-mcp"].)
Cursor / Cline / Windsurf
Point the MCP server command at the same uvx --from git+... steam-mcp invocation and
set STEAM_API_KEY in the environment. See docs/clients.md for
per-client instructions.
Configuration
| Variable | Required | Default | Description |
|---|---|---|---|
STEAM_API_KEY |
for player tools | — | Steam Web API key. |
STEAM_MCP_TIMEOUT |
no | 20 |
HTTP request timeout (seconds). |
STEAM_MCP_LANGUAGE |
no | english |
Language for localized store responses. |
STEAM_MCP_COUNTRY |
no | US |
ISO country code for store pricing. |
Tools
The full, generated tool reference lives in docs/tools.md. Highlights:
- Identity —
resolve_vanity_url,get_player_summary,get_player_bans,get_steam_level - Library —
get_owned_games,get_recently_played_games,get_friend_list - Achievements & stats —
get_player_achievements,get_game_schema,get_global_achievement_percentages,get_current_player_count - Store —
get_app_details,get_app_reviews,search_store - News —
get_news_for_app
Example prompts
Once connected, try asking your assistant:
- "How many people are playing Counter-Strike 2 right now?" →
search_store→get_current_player_count - "What's the rarest achievement in Team Fortress 2?" →
get_global_achievement_percentages - "Summarize the recent reviews for Elden Ring." →
search_store→get_app_reviews - "Whose profile is steamcommunity.com/id/gabelogannewell, and what are they playing?" →
get_player_summary - "What are my 5 most-played games?" (needs your SteamID) →
get_owned_games - "Any recent patch notes for Dota 2?" →
get_news_for_app
Development
git clone https://github.com/rachittshah/steam-mcp
cd steam-mcp
uv venv && source .venv/bin/activate
uv pip install -e ".[dev]"
ruff check . && mypy src && pytest
Contributing
Contributions welcome — see CONTRIBUTING.md.
License
MIT © Rachitt Shah. Not affiliated with or endorsed by Valve Corporation. Steam and the Steam logo are trademarks of Valve Corporation.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。