The Coconut Tutor Census: 77 Cards That Find Your Second Copy

Coconut is singleton: one copy of everything except your named character. If you want to find a specific kind of card reliably, you need cards that search your deck or dig through the top of it. This census covers the full six-ink pool, 2,480 unique cards through Attack of the Vine! with texts as of Sep 6, and lists every effect that takes a card from your own deck based on a filter and gives it to you: to your hand, into play, or to the top of your deck after a full search. There are 77. Scry is not counted: it changes the order of your next draws, but it doesn’t put a card in your hand. Plain draw and loot are not counted, because you don’t choose what you get. Discard recursion is a separate subject and will get its own article. When a filter fits two families, I count the card once, in the biggest family it reaches, and note it in its line.

The math all 77 obey

All these cards follow the same math: the chance that a dig of depth d hits at least one of k legal targets in a 60-card deck is 1 minus C(60-k, d)/C(60, d). The numbers below assume an untouched 60 with no mulligan adjustment, so treat them as minimums.

Three things follow from this table. First: a depth-4 dig finds one specific card 6.7% of the time, so nothing in this census reliably finds a single card. Every card here finds a category, and the useful question is how many cards in your 60 fit its filter. Second: below about 12 to 15 targets, a depth-4 dig hits less than 60% of the time. If your deck can’t reach that count, the slot is usually better spent on one more card of the category itself. Third: for the depth-1 cards, these blind numbers are the worst case, not the intended use. The census has 21 top-1 effects, and they are built to repeat, or to be used after another card has stacked or shown the top of your deck.

Two more numbers before the families. 35 of the 77 are repeatable, on a quest, an exert, or a recurring trigger. That matters more here than in one-on-one Lorcana because four-player games are long: a dig that repeats every turn sees far more cards than an on-play effect of the same depth. And 25 of the 77 are non-inkable (marked * in the tables), so they add up against your non-inkable count fast.

Anything: 12 cards, all Sapphire

The 12 cards with no filter at all are below, and every one of them is Sapphire.

Vision of the Future is the only card in the pool that digs 5 for 2 ink. Sisu — Divine Water Dragon is the family’s only cheap repeatable: depth 2 on every quest, which over a ten-turn game is 8 to 10 filtered looks. Merlin — Intellectual Visionary is the pool’s one unrestricted full search, but only when played with his Shift 5; without another Merlin on the board he is a plain 6-drop, and in singleton your Shift targets are a few one-ofs. Dig a Little Deeper is only realistic when sung. For Coconut this family explains a common pattern: when a deck needs general consistency rather than one card type, the third ink usually ends up Sapphire, because no other ink has any of this.

Characters: 22 cards

These get better the more characters you run. With 30 or more characters in your 60, every depth-4 card here hits 94% or more.

Besties, Assemble! is the cheapest of the depth-4 group, and Be Our Guest is the same effect on a song. The trio in my Stitch deck shows the standard setup: Besties, Assemble!, Nani — Stage Manager and Lucky — The 15th Puppy all point at the same 25 cheap bodies, so all three are well above the 12-to-15 range. Closet Door Portal and Down in New Orleans have the widest filters in the census except the “anything” family, and both put the card they find straight into play. Two cards here are symmetric: Mike Wazowski — Heroic Climber and Let’s Get Dangerous give the same effect to every player, and in a four-player game that means you are giving it to three opponents. I don’t run either without a specific reason.

Tribes: 20 cards

This is where most bad tutor picks happen, so the table carries two extra numbers: how many legal targets exist in the tutor’s own ink or inks (duals included), and in the whole six-ink pool. The in-ink number is what you get if the rest of your deck ignores the tribe; the total is the maximum; your real count comes from the up to three inks you actually run.

Most of the narrow tags don’t reach the 12-to-15 range. Puppy has 12 characters in the whole game and 5 in Sapphire, so even a deck that runs every legal Puppy leaves Television Set at around 10% per blind activation; it needs the top stacked to be worth a slot. Hunny, Red Panda and Dragon have the same problem, with two exceptions: Sisu — Uniting Dragon repeats on every hit, and Christopher Robin — Hunny Sage lets you register Hunny characters from any ink, which also helps Big Book of Hunny in the same list. Two numbers in the in-ink column surprised me. Jim Hawkins — Honorable Pirate is Amber, but only 9 of the 65 Pirates are; Don’t Be Nervous is Amethyst, which has just 6 of the 147 Princesses. Both cards only make sense next to the inks where their targets actually are. The wide tags work fine. Ally covers 662 characters, 205 of them already in Jasmine — Steady Strategist’s two inks, so in practice she is a repeating character-dig. Floodborn covers 183, so Fred — Giant-Sized, the only card in the census with no depth limit, almost never misses. Fred is also 7 ink in Emerald, an ink with almost nothing else in this census, so Emerald’s best selection card usually can’t be played before turn 6.

Names: 6 cards

Singleton hurts this family the most, so the second column counts the copies the name can actually see, in ink and in the pool.

Name-a-card is the extreme case: if you name blind, the flip hits 1.7% of the time in a 60-card deck, and nobody plays these blind. The Sorcerer’s Hat and Bruno Madrigal — Undetected Uncle are meant to be used when you know the top card: another effect puts the card there, or shows it to you, and then the naming always hits, with Bruno adding 3 lore every time. The fixed names have a different problem: few copies exist. Charles Muntz — Obsessive Explorer is the clearest case. He is Sapphire, and the game’s only Kevin is an Emerald card, so his quest trigger does nothing unless Emerald is one of your inks and that one Kevin is in your 60. Focused Search’s Kevin clause has the same problem, but its item half still works. Pete — Wrestling Champ sees 3 Petes inside Steel, Yzma — On Edge needs three specific cards in the right places, and The Queen — Crown of the Council can take several Queens at once from a name with 14 printings. One rule helps the whole family: “named X” matches every version of X, so when X is your coconut’s named character, four of your legal targets are the same card and the tutor becomes very reliable.

Items: 9 cards

Item filters only work if you run enough items: with 10 to 14 in your list, the depth-4 cards here hit 53 to 66% each. Below 10 items, skip the family. Same pool columns as above.

Eight of the nine are Sapphire, and Sapphire also has 51 of the game’s 205 items, a quarter of the type in one ink, so the tutors and their targets are in the same ink. The repeatables are the ones to look at first. Basil’s Magnifying Glass digs 3 every turn for 2 ink. Hiro Hamada — Robotics Prodigy searches the whole deck on a repeatable activation and puts the card on top, so it sets up your next draw rather than this turn’s play. The Queen — Diviner plays cheap items for free and puts the expensive ones in your hand. Oswald — The Lucky Rabbit reveals the top card every time you ink during your turn, and plays it for free if it’s an item. My Adventure Book actually reads “non-character”, which is 104 cards in Emerald alone; I count it here because item decks are the ones that run it.

Locations: 3 cards

The shortest list in the census. All three cards are Ruby, an ink with 18 of the game’s 85 locations.

The Islands I Pulled from the Sea always finds one. Amazu’s Inkcaster is the repeatable. The Cold Never Bothered Me is the fast version and has a real cost: the three misses go to your discard, and in singleton those cards are gone for the game. The conclusion: a deck that wants to search for its locations, instead of just drawing them, needs Ruby, because all three of these cards are Ruby.

Songs and actions: 5 cards

The smallest family. Four of the five put the card straight into play instead of your hand.

Ariel — Spectacular Singer is the only one that puts the song in your hand, with 28 Amber songs available before you count your other inks. Powerline — World’s Greatest Rock Star and Robin Hood — Sharpshooter both dig 4 and play the card they find for free, and both are Ruby. Robin Hood’s leftovers go to the discard, same warning as above. Prophetic Vision has the census’s only “each opponent” text, and it triggers on the miss: if the flip fails, each opponent loses 1 lore and you gain 1.

When the tutor plays it for free

Twelve of the 77 put the card they find straight into play, plus The Queen — Diviner when the item is cheap. For these, the hit chance is only half of the evaluation; the value of what you play for free matters just as much. The depth-3-plus versions (Powerline, Robin Hood — Sharpshooter, Closet Door Portal, Down in New Orleans) work on their own. The depth-1 versions are meant to be set up, and some setups are in the census itself: Hiro Hamada — Robotics Prodigy puts an item on top and Oswald — The Lucky Rabbit plays it for free on your next inked card, both Sapphire, and Gazelle — Ballad Singer puts a song from your discard on top for Mulan — Reflecting to play on her next quest. Without setup, they are a bonus on a character you wanted to play anyway, which is how Mufasa — Betrayed Leader, Pete — Wrestling Champ and Chief Bogo — Commanding Officer usually get used.

The other detail worth checking on any top-1 card is where a miss goes. To the bottom means the card filtered even when it missed, so it is safe to use blind: Pongo — Determined Father, Monterey Jack — Watchful Ranger, John Smith’s Compass, Television Set, Big Book of Hunny, My Adventure Book and Oswald. Back on top means a blind miss changes nothing, so those (both Sensor Cores, The Sorcerer’s Hat, Bruno Madrigal — Undetected Uncle, Mulan — Reflecting) are the ones to use when you already know or control the top.

The census on one screen

Rows show what the card finds; columns show how. “Rep” counts the repeatables inside each cell.

By ink, counting duals on both sides: Sapphire 33, Amber 28, Amethyst 8, Ruby 6, Steel 4, Emerald 3. Sapphire has all 12 unrestricted filters and most of the item tutors. Amber has most of the character tutors. Ruby’s entries mostly play the card instead of putting it in hand. Amethyst has few entries but they combine with each other: its top-of-deck peeks set up its own name-a-card payoffs. Emerald and Steel have almost nothing, so decks in those inks get their selection from a second ink or play without it.

What did I miss? Card texts are as of Sep 6, 2026, and the app rewrites something almost weekly, so if a line above stopped being true, tell me which one.