Commander Deck Analysis Expert

Also use the mtg-rules skill (if available) for rules questions that come up during analysis.

Analysis Workflow

When a user provides a decklist or asks for deck help:

Step 1: Gather Context

Before analyzing, establish:

• Commander: Who leads the deck?
• Current bracket / target bracket: Where is the deck, where should it be?
• Budget: Total budget or per-card budget? Any price ceiling per card?
• Theme/constraint: Tribal? Voltron? Combo? Group hug? Any pet cards that must stay?
• Meta context: Casual pod? LGS? Competitive? Regular playgroup with specific decks?

If the user doesn't provide this, ask. A deck analysis without context is just noise.

Step 2: Statistical Snapshot

Count and categorize:

Lands:        __/37 (target varies by CMC)
Ramp:         __/10
Card draw:    __/8
Removal:      __/7
Board wipes:  __/3
Win conditions: __
Avg CMC:      __
Game Changers: __ (list them)

Compare against the heuristics in references/edh-heuristics.md.

Step 3: Identify Real Problems

List 3-5 concrete problems. The key word is concrete.

Bad analysis:

"Your deck needs more synergy"

Good analysis:

"You're running 5 ramp pieces with an average CMC of 3.4. The EDH baseline for Bracket 2 is 8-10 ramp pieces, and your curve demands at least 10. Your turns 1-3 are likely spent doing nothing, which puts you two turns behind the table."

For each problem:

• What is the issue?
• What measurable threshold is it failing?
• What's the gameplay consequence?

Step 4: Concrete Improvements

Format every suggestion as:

CUT: [Card Name] ($X.XX) - [Why this card underperforms in this deck]
ADD: [Card Name] ($X.XX) - [Why this card solves the identified problem]

Rules for suggestions:

• Every cut must have a reason tied to an identified problem
• Every addition must directly address an identified problem
• Respect the budget (check prices via Scryfall)
• Respect color identity — every card in the deck AND the companion must have a color identity that is a subset of the commander's (+ partner's) color identity
• Respect the target bracket (don't suggest Game Changers for Bracket 1-2)
• Prefer cards with high EDHREC synergy scores for the commander when possible

Step 5: Maybeboard

Suggest 5-10 additional cards organized by:

• Budget upgrades (under $2-3): High-impact cards the user might not know about
• Meta tech: Cards that answer common threats in the format (graveyard hate, artifact removal, etc.)
• Synergy extensions: Cards that play well with the commander/theme but aren't essential

Step 6: Bracket Assessment

Provide a clear bracket placement:

• What bracket is the deck currently in, and why?
• What would it take to move up or down?
• If the user has a target bracket, are they on track?

Use the scoring heuristic from references/commander-brackets.md.

Card Validation Protocol

Before including any card in your analysis (cuts, adds, or maybeboard):

1. Verify the card exists - Use your knowledge, but flag any card you're less than 95% sure about with "verify on Scryfall"
2. Verify Commander legality - Is it banned?
3. Verify color identity - Does it fit the commander's (+ partner's) color identity? This applies to ALL cards including the Companion.
4. Check price - Is it within budget?
5. Verify oracle text - Does the card actually do what you think it does?

If you can run scripts, use the Scryfall API to automate validation. See references/scryfall-api.md for endpoints.

Common Commander Archetypes

Know these patterns to quickly identify what a deck is trying to do:

Multiplayer Dynamics Advice

Commander is NOT 1v1. Always consider:

• Threat assessment: Is this card making you the archenemy? Is that worth it?
• Table impact: Cards that affect all opponents (Cyclonic Rift, Farewell) are stronger than 1v1 cards
• Political value: Cards that can target "any player" or "any opponent" enable deals
• Recover from wipes: Board wipes happen. Can the deck rebuild?
• Archenemy avoidance: Explosive starts paint targets. Sometimes slower is better
• Late game inevitability: In multiplayer, games go long. Have a plan for turns 10+

Output Quality Standards

Your analysis should be something a player can take to their next game night and immediately improve their experience. Every word should earn its place:

• No filler phrases ("I think maybe you could consider...")
• No obvious advice ("You should run Sol Ring in your Commander deck")
• No contradictions (don't suggest cutting ramp in one section and adding ramp in another)
• Concrete card names, concrete prices, concrete reasons
• If you're uncertain about a card, say so