steam-mcp
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
No arguments | |||
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": true
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| recommend_gamesA | THE MAIN TOOL. Recommend games matching a stated preference, ranked against what the player actually plays. Describe the mood in |
| get_taste_profileA | Derive what a player actually likes from their library: tag affinities weighted by playtime, top genres, signature games, and titles they bounced off. Use this to explain or sanity-check a recommendation, or to answer 'what kind of games am I into'. Requires a Steam API key. |
| search_gamesA | Search the Steam store with structured filters and return full details for each hit. Use when you want raw discovery without any personalised ranking; recommend_games is usually the better starting point. Keyless. |
| get_game_detailsA | Full detail for one or more games by appid: description, genres, community tags, price, reviews, Metacritic, platforms, estimated owners, and optionally Steam Deck compatibility. Keyless. |
| find_similar_gamesA | Find games resembling one or more you name: 'more like Hades', 'something between Factorio and Rimworld'. Takes the seed games' dominant community tags and searches on them, scoring hits by tag overlap. Keyless, but personalises when a key is available. |
| get_game_reviewsA | Score breakdown plus the text of the most helpful reviews for one game. Use this to judge whether a game actually delivers on a specific preference (pacing, difficulty, story, how well it runs), which tags alone cannot tell you. Keyless. |
| list_tagsA | Search Steam's official tag vocabulary. Call this before recommend_games or search_games when you are unsure a tag exists or what it is called: 'roguelite' and 'Rogue-lite' are different strings and only one is real. Keyless. |
| get_my_libraryA | The player's owned games with hours played, most-played first. Requires a Steam API key and a profile whose game details are public. |
| get_wishlistA | The player's Steam wishlist, with current price and review data for each entry. A strong signal of stated intent, and useful for spotting what is on sale right now. Works without a key when the profile is public. |
| get_recently_playedA | What the player has actually played in the last two weeks. The best read on current mood, as opposed to the lifetime habits get_taste_profile describes. Requires a Steam API key. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 10 tools
Each tool targets a distinct action and data source: taste profiling, recommendation, raw search, details, reviews, similar games, tags, and player-specific library/wishlist/recent activity. Even the three discovery tools are clearly separated by input type (preference statement, structured filters, seed games) and output purpose.
All tool names follow a consistent lower_snake_case verb_noun pattern, with get_ as the dominant verb and action-specific verbs like recommend, search, find, and list used uniformly. There is no mixing of conventions or vague generic names.
The 10 tools are well-scoped for a Steam-focused MCP server: they cover discovery, personalization, game details, reviews, and player data without unnecessary duplication or bloat. Each tool earns its place in the workflow.
The tool surface forms a complete read-only pipeline: understand the player, query the catalog, refine by tags, get details/reviews, and generate or sanity-check recommendations. The only possible gaps (e.g. account actions) are outside the server's apparent purpose.