Ask your agent about Magic: The Gathering cards and your own decks in ArcaneMetriX: card roles, synergies, commander brackets, and the cards worth adding.
Scored across 10 tools
Three tools converge on bracket checking: analyze_my_deck (saved deck), bracket_check (raw decklist), and deck_analyze (decklist with bracket plus themes/categories). An agent could reasonably hesitate between bracket_check and deck_analyze, and analyze_my_deck overlaps with get_my_deck's domain. Descriptions do clarify scope (saved vs. decklist), which keeps it from scoring lower.
All snake_case, but the verb/noun pattern is mixed: some leading-verb (analyze_my_deck, check_my_collection, get_my_deck, list_my_decks, suggest_cards) and some leading-noun (bracket_check, card_lookup, card_synergy, deck_analyze, my_pod_schedule). The 'my' qualifier floats inconsistently (analyze_my_deck vs. my_pod_schedule), though names remain readable.
Ten tools for a focused MTG deck-analysis server is well-scoped. Each covers a distinct need (lookup, synergy, collection, deck read, bracket, suggestions, schedule), and none feel redundant padding.
The read/analysis surface is broad and coherent: cards, synergy, collection, decks, brackets, suggestions, pods. The main gap is deck lifecycle—there is no create/update/delete or add-cards-to-deck operation despite being able to read decks, leaving agents unable to persist changes.
The bracket check of one of the user's saved decks: its bracket and the reasons behind it. Needs decks:read.
Inputs: deck_id (required).
A Commander decklist's bracket (1-5) and the reasons behind it: Game Changers, two-card combos, mass land denial, extra turns and tutors, each with the cards involved.
Inputs: decklist (required).
A Magic card's identity (mana cost, type, color identity, legalities, image) and what ArcaneMetriX reads in it: its role in a deck and its tags. Takes a card name or a Scryfall oracle id; a name that matches nothing exactly returns close matches to choose from.
Inputs: card (required).
The cards that work best with a card, ranked, each with what they share. Give with to ask instead whether two particular cards work together, how strongly, and why.
Inputs: card (required), with, limit.
How many copies of each card the user owns in their ArcaneMetriX collection (needs their PRO membership and collection:read). Takes up to 100 card names or oracle ids.
Inputs: cards (required).
A Commander decklist read whole: its bracket and why, the themes it is built around, and the functional categories (ramp, card draw, removal, …) it is short on.
Inputs: decklist (required).
One of the user's decks card by card, with its commanders, sideboard and notes. Needs decks:read.
Inputs: deck_id (required).
The signed-in user's ArcaneMetriX decks, newest first: name, format, commanders, card count and bracket. Needs the user to connect their account (decks:read).
Inputs: limit, offset.
The upcoming game nights the user's pods have planned (a user may be in up to three): which pod, when, where, what formats, how many are going and the user's own answer. Pass pod_id for one pod's. Needs pods:read.
Inputs: pod_id.
Up to ten cards to add to a Commander decklist, ranked, each with the reason, built toward a bracket when one is given.
Inputs: decklist (required), limit, bracket.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Add one secure layer between your agents and this server.