Nomad Cairn

Tested by breaking it on purpose

0 duplicated · 0 lost

across 160,000 items and 2,500 simulated death-and-retrieval cycles with 5 players, every item counted at the end.

• 859 automated tests. 0 failures.
• 24 crash-and-recovery stress tests, the server killed mid-retrieval on purpose. 0 failures.
• 190 live-server cases on Folia 26.2, plus 190 on Paper 26.2. 0 failures.
• 160,000 items in the simulated run, each accounted for exactly once: 130,848 handed back, 29,152 still safely held.

Transactional custody ledger

Every item is booked in when a player dies and booked out when it is handed back, like a bank ledger. An item is never in two places, or none.

Crash-injected

The tests pull the plug at chosen points on purpose, then check where every item ended up.

Partial delivery

A full inventory takes what fits. The rest stays safely held in the ledger until there is room.

Fast-restart recovery

When the server comes back quickly after a crash, Cairn works out from its records which side of the hand-over each item is on.

The volume run is a simulation rather than a live server, and the results describe Cairn's own accounting. Read every scenario, including what the suite does not prove.

The guarantee

Cairn tracks every item a death puts in its custody in a transactional ledger. If the server dies in the middle of a retrieval, the hand-over resolves to one side: the player has the items or the ledger still holds them, never both and never neither. That is a claim about Cairn's accounting, not a promise that a crash costs nothing. Minecraft writes a connected player's inventory on quit and at a clean shutdown, not continuously, so a hard kill can take items that were never on disk, below Cairn and below every other plugin. Anything Cairn cannot decide after a crash is held, reported, and handed back with /cairn restore.

The guarantee covers items in Cairn's custody. A rollback plugin, an inventory editor or a second graves plugin reacting to the same death can create items outside it, so Cairn checks for another graves plugin at startup and warns if it finds one.

The evidence

Every release runs a stress suite that crashes a server at chosen points in the death and retrieval paths, then counts every item: each one must be in an inventory or in the ledger, exactly once. It also runs against a real Folia server with headless clients. The figures below come from the run that produced the current jar.

Run on 2026-09-23. Read every scenario, including what the suite does not prove.

What it does

• Six marker styles. Player head, armour stand, block display, hologram only, headstone, and corpse (a life-size figure lying where the player died, wearing their skin). Set a default per world, or let players pick their own with /cairn style, one permission node per style.
• Regions. With WorldGuard, a cairn-death flag gives a region its own rule: no-marker keeps blocks out of a build while still holding the items, and keep-inventory, drop, relocate and normal cover the rest. Without WorldGuard, plain coordinate zones in config.yml do the same.
• Fees and permission tiers. Grave count, expiry time, XP retention, teleport and fee waivers are all set by permission, so better death recovery can be a rank perk. An optional Vault retrieval fee, flat or by distance, is skipped rather than blocking a player from their own items if no economy is installed.
• Grave compass. An ordinary compass a player carries points at their nearest grave in the current world, or their most recent one elsewhere. No item is given.
• Join reminder. On login, a player is told in chat how many graves are waiting for them.
• Hologram. The label above a grave shows the owner, a summary of the contents, and a live countdown of the protection window that disappears when it closes. Blank a line in messages.yml to hide it.
• /cairn restore. Rebuilds a grave from the audit log where the admin is standing. It refuses if the ledger says the items were already delivered; an admin can override that, and the override is logged under their name.
• /cairn verify. Cross-checks every grave against the ledger and reports anything to look at, without changing anything. The same check also runs on a timer and when a player joins.
• /cairn history. Every change to a player's graves, with a time and a reason, so a missing-items ticket starts from a record instead of a guess.

Not supported: Spigot and CraftBukkit

Cairn will not load on Spigot or CraftBukkit. Its crash safety depends on writing a delivery record into the player's persistent data container, which Paper saves in the same file as the inventory, so the two survive or roll back together. That is how Cairn answers "did the items reach them?" after an unclean shutdown. Spigot has no equivalent, and a Spigot build would be a weaker product under the same name. Check your server software before buying.

Install

1. Put the jar in your server's plugins/ folder.
2. Start or restart the server once. It writes config.yml and messages.yml to plugins/NomadCairn/, and the defaults work unchanged.
3. Optional: install Vault and an economy plugin and set retrieval.fee.model for retrieval fees, or apply the LuckPerms rank template for tiered limits.
4. After editing the config, run /cairn reload. Moving storage.folder needs a restart.

To check it works, die holding a few items. A player head appears where you died; right-click it or run /cairn to get them back.

Before you buy

• Version history: changelog.
• Test evidence: every stress scenario and its result.
• SHA-256 of NomadCairn.jar inside the download, identical for every buyer (the download page also shows the zip's own hash): 3674af36bf3f9b159ec7c2e538bc496d45b48c5eba9cb3ce40ae34f0b83ac72b
• Licence: the Commercial Plugin EULA. Refunds: refund policy.
• Support: nomad-labs.co/support. Cairn is maintained by one person. Include the output of /cairn verify and the log lines around any error.

Questions