Update 5.5 "Deck Drop" is LIVE (2026-08-31). Two independent sweeps confirm the public-branch release (Channel37's trailer: "STATUS: DEPLOYED... the seventh Early Access update"); the ShadowPC install read buildid 25056839 first-hand on 2026-09-01 (0.8.5.650920, engine reported 5.8.1 on the beta). The corpus docs still carry the 0.8.0.612251 anchor -- the next-patch batch (shadow-install re-inventory, full re-mine, re-anchor docs 01/03, lift the caveats) is pending Robert's go; every [beta-era] tag now reads as 5.5-era, and the datamined human-growth values in doc 23 are already 5.5 values.

This is the mandatory session pre-read. It is the working fluency layer; the 22 numbered docs in docs/ are the detail layer (routing table in §14). Ground truth for install/engine/save facts: docs/ref/shadow-install.md (first-hand, trusts over any web source). The numbers layer for power and fluids: docs/ref/flow-research/ (five reports) synthesized into doc 10.

0. Spoiler contract (BINDING for every companion session)#

Robert started playing 2026-08-24. As of 2026-09-05 he has restored the Lazarus Complex, racked the first seed, grown a human, flown the first launch ("Ascension"), opened all three Eden Cradle chambers, built Black Vein Station into his oil / diesel / steel / carbon-fibre station and made its first batch, built the first Eden Frame at Pishon and planted the first pear, and grown and launched a full Habitat Care group (four seeds from one Prometheus run, fed to the cookbook's recipes) that is in orbit waiting for the committee -- he is in the committee / Eden / industry era. As of 2026-09-07 he is level 32, has racked a seed for Project Jonah at Platform Omega and is wrangling sharks for the Omega Nursery pens, and plays on a 15-container mod stack (docs/mods/18-installed-mod-stack.md: ten Nexus quality-of-life paks plus the house Matte Black livery, an Enhancer Cap of 150, x10 cables, a 150 kg net, x10 boxes with a 400 m crafting pull) -- every number in this corpus is vanilla unless a doc says otherwise. He remains spoiler-averse for everything he has not yet reached; the contract below is unchanged. (Playthrough state otherwise lives in praxium, never in docs/ -- this status line is the one exception, kept for spoiler calibration.)

1. Everything above the SPOILER WALL (§S1+) in this doc is mechanics-grade or officially-marketed information -- safe to use freely in answers.
2. Everything below the wall is story: never volunteer it. If Robert asks a question whose answer requires fenced content, say so first ("that crosses into story -- want it?") and reveal only on his explicit yes.
3. "What should I do next" answers default to spoiler-light routing: mechanics, preparation, direction -- not quest outcomes, not what a location contains, not who a voice is.
4. Docs 04 (lore), 17 (quests), 18 (humans), 20 (achievements) are spoiler-dense; quote from them carefully. Docs 05-13, 19, 21, 22 are mostly safe; doc 11's late sections and doc 12's reveal-map are mild-spoiler.

1. The game in one paragraph (spoiler-free)#

You are the last functioning Caretaker, a humanoid robot on an Earth drowned by a global flood, alone on an open ocean dotted with half-sunken facilities. Your standing mission (openly marketed): recover human seeds, restore the Lazarus Complex, grow humans, and launch them to orbit -- humanity survives only if you keep working. Mechanically it is a first-person survival-logistics game where the survival meter is electric charge, not hunger: you sail a 30-metre boat that is simultaneously your vehicle, base, power grid and workshop; you salvage finite wrecks, restore power to dead facilities by physically running cables, craft through a proximity-based fabricator, fight machine and bio-machine hostiles with real firearms, and push progression through a skill tree + body-upgrade system. Single-player only. Closer to Raft than Subnautica in structure -- surface-first, waypoint ocean, quest-driven -- but its systems are its own.

2. The version story (know this before trusting ANY guide)#

Decoder facts: marketing numbers don't track semantic versions -- the 4th field (a monotonic CI counter, ~39/day) is the only reliable ordering key. Channel37 pushes the depot day N, posts land N+1 (depot dates authoritative). Since U05 there's an in-game patch-notes reader (Main Menu -> Patch Notes) -- first-party and better than any URL. Directional balance notes only ("reduced", "improved") -- almost no published numbers, and no developer has ever published a power figure (every kW in circulation is tooltip- or player-derived). Public betas ship as a Steam branch behind a published access code (hence the "Latest Beta Code?" threads).

Stale-guide tripwires (launch-era claims now FALSE): max level 30 · no photo mode · no flying vehicle · storage placement permanent · boat speed bugged · vehicles too expensive to run · 8 enhancers (now 11 -- Equipment Handling is the undocumented eleventh) · 21 skill tiers (a v0.5.0 snapshot; 25 is current) · hardcoded gyro keys (customizable since U05) · CS-17 Skimmer at T13 (datamined T9) · wind turbines at 1-2 kW (a wiki field misread; 5-10 kW) · "carry oil for range" (U05 halved the oil-to-diesel yield). Anything dated before 2026-06-24 is at least one major update stale.

Roadmap state: original six blocks, five shipped (3.5 was a bonus). Block 6 = "The Future / New Frontiers" -- unannounced. No 1.0 date exists ("full launch in 2026" was a launch-day aspiration never restated). The "12 months of EA" elapses 2026-11-06.

3. Systems map -- how everything interlocks#

  SALVAGE (finite world objects, wrecks, POIs)          RENEWABLES (boat net trickle,
      | Dismantle Tool (T2)                              respawning floaters, enemy drops,
      v                                                  cocoon ranching, farming since U05)
  SCRAP (11 types) --Recycler--> MATERIALS (4 rarity bands)
      |                              |
      |                              v
      |                    FABRICATOR (10 m radius, 3x hand speed, queues)
      |                              |
      v                              v
  XP (all three verbs pay XP)   PARTS -> MODULES / TOOLS / WEAPONS / VEHICLES
      |                              |
      v                              v
  LEVEL (cap 35) --+--> SKILL POINTS -> 25-tier recipe tree
                   +--> ENHANCER POINTS -> 11 body stats x 100 levels
                                     |
  POWER: generation (solar/wind/fuel) -> batteries -> cables (20 m segments, 4 families)
      -> boat bus (6 kWh + 14 outlets) -> facility yellow switchboards -> YOUR OWN BATTERY
                                     |
  BOAT = vehicle + base + grid + workshop + respawn/backup + storage + net
                                     |
  ENDGAME LOOP (mechanical): seeds -> Lazarus Racks (power+water) -> humans -> launches
      -> orbit -> capsule drops (farming) -> committees

The load-bearing insight: electricity is the one currency -- your body, your tools, your boat, every facility and the endgame all draw from the same watt economy, and moving charge (cables, sockets, backpacks) is the actual gameplay. The second insight: the world is finite but the game is not soft-lockable -- fixed POIs deplete permanently, while the boat net, floating containers, enemy respawns, cocoon colonies and farming provide an infinite (slow) floor.

4. The numbers that matter (all verified current at 0.8.0 unless tagged)#

5. Opening playbook (the first-session cheat sheet)#

The opening chain is fixed: From Darkness, Light (battery -> outlet -> main switch) -> Where Waters Wait (docking hangar; the Heal Station and Backup Station objectives are optional -- do them anyway) -> Light the Way Home -> Vessel of Tomorrow (wake the boat; 4 of its 5 tasks are optional, only "open dock doors and sail out" is required -- do all five, sailing out unprepared is the classic self-inflicted wound) -> Rags to Riches (Recycler) -> Voyage into the Unknown (Navigation Beacon 49) -> First Light (fabricate tutorial).

The known early blocker: a glowing creature guards a door before you have any weapon. The answer is melee -- the C key (a top Steam-thread stall for new players).

Doctrine, in priority order:

6. Hazards & combat quick reference#

Meters: battery and vitality, nothing else. No hunger/thirst/oxygen/temperature/radiation/toxins -- verified absences. Damage sources are exactly: enemies, falls, battery-empty drain, over-depth pressure.

Death = a save reload, dressed in fiction [settled first-hand 2026-08-27]. "Terminated -> return to a backup" literally rolls back to your last autosave or manual save -- no item drop, no corpse run; you lose only the play time since the save (autosave = 10-min cadence). Practical rule: manual-save before every risky poke (new outpost, dive, night landing) -- it converts death from a setback into a 30-second rewind. Corollary: the rollback undoes ANYTHING since the save, which is also the emergency brake on mistakes.

Enemy families (17 catalogued; full roster + tactics in doc 06): cocoon-spawned bio-swarms (slimes, crawlers, spitters), rogue machines (Roller Boi, Laser Bob), cybernetic sharks (Ram/Talon/Laser), floating Angel/ArchAngel gun platforms (Angels are passive until provoked -- seeing a carried seed or human counts as provocation; they throw slow blue orbs that walls block; never shoot the ones over your own complex; first-hand time-to-kill: ~4-6 Bolt Sniper rounds each. The ArchAngel is the cannon variant: attacks on sight at range, volley-then-reload, drops Bio Flesh 50% -- fight it from cover, ideally the boat deck with the Heal Station behind you). Cocoons are the spawn source and they regrow from intact ROOTS, in the dark first-hand 2026-09-01: destroying the cocoon body is temporary, destroying the roots is the permanent clear -- which is why "cleared" POIs repopulate. Tentacle contact damages you (confirmed). Roller Boi kill-chain: shoot the cannons off -> back off -> rifle the body from range (it explodes on death like a grenade; disarmed units flee the dismantle tool). Laser Shark: the laser is dorsal but concealed, deploying upward before it fires (a telegraph); it drops ~2x a Talon's loot -- shark hunting is the renewable rare-scrap channel. Ammo doctrine: Slam rounds for sharks, Tri rounds for machines.

The attractant economy is the tactics layer: electricity draws Sweethearts · petrol draws Mosquitos (NEVER fight them with fire) · light draws slimes and (since U05) aquatic swarms · sound draws Angels · dropped organics get eaten by scavenging crawlers. Black Vein Station first-hand 2026-09-03: the Angels were the easy part -- the gate was Mosquito swarms from cocoons on the helipad above the dock, unreachable from the boat, with the light-loving slime on the lights and the dark-loving slime in the dark at the same time (in the 5.5 art the light-lover looks red and the dark-lover black -- the wiki's "Pink" and "Red" names are the other way round); what worked was strip to health-and-offense, sprint the base for the nest, and carry a battery for the dead elevator (doc 06 §7.11). Light is also a weapon: night crawlers and red slimes die in intense light or daylight, and the Eclipse headlamp (T18, ~10 kW) kills night crawlers on contact -- light walls plus a lit cocoon colony make a ranchable bio source (doc 06).

Weapons: five families -- Melee, Bolt (firearms), Electric, Flame, Thrown. HARD RULE: only Bolt weapons work underwater. Bolt Pistol T5, Rifle T11, Sniper T17. Electricity backpacks double as energy-weapon magazines (and boost electric-weapon damage).

No lightning-rod mechanic is obtainable (its skill is a "Never" entry; a description string exists -- "protects surrounding modules", nothing about bonus energy). No base raids, no ship turrets; defense is positioning, floodlights (keep leeches off the deck), and the U03 companion units (FELIS, tamed Sweetheart).

7. Vehicles#

Boat damage: storms > mines > collisions > ramming enemies. Sinking exists; standing fix "repair, raise net, raise anchors." No autopilot; the boat drifts slowly to port unattended. Electric-only cruising: one 6 kWh battery buys ~3.5 km of hard running at ~20 km/h; a fully boosted electric drive draws 28.5 kW at full throttle (about one Large generator).

8. Power doctrine#

One electrical currency, several un-networked pools (your body, boat bus, standalone batteries, per-facility wiring) -- moving charge is always a physical cable run. Socket modes via F: output / input / equalize (the default, and the only "grid logic" in the game). Equalize levels ABSOLUTE charge per unit, not percentage -- small full batteries pin a big bank at their own Wh level forever; filling a large store past that equilibrium requires Input mode on its connection. Ports are connection-classed (generator outputs cannot link to each other -- route every generator into a battery, extender, boat socket or wall socket); cable bandwidth ~1 MW = never the bottleneck; a per-port ceiling is refuted (60+ kW measured through single ports and extenders). A generator gauge reads connected LOAD, not output -- an unfull Medium Battery is what makes a whole fleet register. Solar = LUMI series [SETTLED 2026-08-29, tooltip + mine agree]: LumiLite 3 kW (T2) / LumiMid 6 kW (T6) / LumiMax 10 kW (T13) peak, gimbal-tracked -- the wiki's 900/4,000/8,000 "Wh optimal" figures are retired. Wind [same evidence] = Whisper 2.5 kW (T3) / Cyclone 5 kW (T8) / Typhoon 10 kW (T14, storm-rated: storms are a power EVENT -- wind spikes above nameplate while solar dies; hybrid generation is the answer). The Typhoon's two ports are a 50/50 circuit splitter (one cable connected = the full 10 kW; both = 5+5 kW to two separate circuits; turbine-to-turbine is refused). Turbines decay (~10%/day launch-era, reduced in 3.5; a Typhoon needs repair on every revisit) while solar has zero upkeep -- hence LumiMax as the daily driver. Fuel = Portable 10 kW (T5) / Large 30 kW (T15), demand-driven (they throttle to the downstream draw); their CO2 output is optional to connect (feeds Sabatier/carbon-fiber chains) but a FULL downstream CO2 tank stops the generator -- leave it unconnected for uptime until you need the chain, and when you do, the Sabatier reactor (60 kW; 60,000 L/h CO2 -> 60,000 L/h methane, plus 5,000 L/h FRESH water -> 120,000 L/h O2 [datamined pass 2] -- feed it purified water, not the salt line) self-relieves through its spare oxygen ports and throttles its own conversion when an output backs up first-hand 2026-09-01. Grid-sizing traps: the Recycler's running draw is contested (REPORTED ~90 kW vs 200 W in the module data -- §4); headlamps are your biggest personal draw (Eclipse ~10 kW = 40x idle); a single low-condition battery caps a whole bank; dismantling a battery deletes its charge.

Meters: boat dashboard = pooled built storage + hull internals; PECO Grid Scan 2000 (T15, 2 SP) = built modules only, by name, with live watts and switch states -- the authoritative infrastructure audit (it also finds the dead hop after a reload). The 0.8.0-era "Lazarus reads 122 kWh with ~62 unaccounted" puzzle is retired: on the beta the scanner accounts every Wh as built (no native facility storage), so treat it as a historical artifact, not a mechanic.

Facility restoration is not a puzzle: every dead structure has one yellow switchboard -- cable power in and the whole site (lights, doors, elevators, terminals) wakes. Repair the site's own rooftop panels/turbines, or run cable up from the boat mooring (extender per floor -- the 20 m rule). Shortage = brownout (things slow, nothing breaks); live loads beat storage charging; no priority config exists. Facility wall receptacles (and the buildable U05 wall sockets, T15-18) are paired pass-throughs, not outlets -- two per wall means two independent lines may cross it. A Lazarus rack that says "No Power" may actually lack fresh water (dev troubleshooting doc); keep salt-water and fresh-water networks separate. Mega-sites (Exodus) need generation and storage in each area -- distant cells unload. Eden Cradle has two grids: the top building feeds the upper floors and the elevator, and the tunnels below have their own inlet console -- run cable down the staircases, not the shaft (doc 12 §7.10.8). Remote production sites (Black Vein: pump ~60 kW reported + tower 60 kW) outrun the boat bank in ~40 min -- bring generation, LumiMax not Typhoon, and a kWh bank sized for the batch (doc 10 §9.6).

Consumable batteries (25/100/200 Wh) are emergency snacks against a 500-2,500 Wh tank -- carry a couple, never rely on them. Full treatment: doc 10 v2.0.0 and the five reports under docs/ref/flow-research/.

9. Crafting & economy#

The three-verb loop: Dismantle (world -> scrap; T2 tool; high-durability targets refund the full parts list; the preview shows yield + Wh cost before you commit) -> Recycle (scrap -> materials; organics are the ONE class the Recycler refuses -- they go to the Soilforge or the Diesel / Petrol Refineries; parts -> recycler -> materials round-trips ~1:1) -> Fabricate (materials -> parts -> everything). All three verbs pay XP -- industry IS leveling. Dismantle-then-recycle equals direct recycling on materials but pays XP twice.

Rarity bands: Common (Copper, Iron, Glass, Plastic, Rubber, Fabric) · Uncommon (Aluminium, Gold, Polyurethane, Silicon) · Rare (Carbon Fiber, Titanium, Stainless Steel [U05] -- not a skill: a Fabricator recipe that appears when an Oil Tank is cabled into the oil port under the Fabricator's panel, 1 Iron + 10 L oil each; an Eden Frame costs 50) · Artifact (Neutronium, Plasma Gel, Quantum Crystal). Hoard Titanium/Carbon Fiber/artifact-tier; stop hauling copper/gold once established. Ore is mineral clusters on the spires and their underwater bases, concentrated in the far north-east Quarantine Stacks; find it by eye from a height (sonar does not mark ore), shoot down what you cannot reach, keep Quantum Crystal as nuggets (doc 12 §7.8, doc 09 §2.3). Build forward Recyclers at salvage sites rather than hauling scrap home. Security-gated rooms hold the top-band scrap density (derelict "Aged Lazarus" vessels dismantle into Ultra Tech Scrap first-hand).

Renewables that actually renew: shark hunting (Laser Shark ~2x loot), ranched cocoon colonies for bio waste, the boat net, floating containers, and -- after the first launch -- the Farming Bed, which eats clean water plus large amounts of bio waste dropped on it as soil. Hoard biowaste from day one; it is soil later.

No conveyor/belt automation exists -- "logistics" means containers, tanks, cables, the forklift and the boat. Unattended converters (Recycler, Soilforge, generators, purifiers, Sabatier, Distillation Tower, U05's Recovery Grinder (T23, 4 SP) = the late-game material transmuter that converts the infinite low-grade stream into depleted rares at an oil cost) are the automation ceiling. Production enhancer modules are 20-charge consumables that attach to a machine (max 2 slots, both deplete per craft) -- attach only before expensive batches.

Building is module placement, not construction: no foundations/walls/terrain; ~54 modules snap to existing surfaces (boat deck, facility floors); green/red ghost; B build menu confirmed first-hand; dismantle = full parts refund (dev-confirmed) scaled by durability -- repair before scrapping. Land "bases" = occupying existing POIs. Storage containers relocate freely since U05; everything else is dismantle-and-rebuild.

10. Progression & the companion protocol#

Two currencies per level-up: skill points (recipe tiers) + enhancer points (body). Almost nothing is story-locked -- the real gates are fuel range, dive depth, carry weight, power, and map knowledge (Navigation Beacons reveal POI clusters; beacons chain the world open). Growing for a seat: doc 23 is the cookbook -- a feed recipe per Council seat computed from the datamined tables (the wiki's profession thresholds are wrong in 21 cells; a gated attribute must reach its threshold (the data tags the check Greater, but two rack screens read 100% with an attribute exactly at its gate first-hand 2026-09-05 -- the recipes keep a +1 margin as insurance); selection takes the highest tier you clear in each family, then rolls among families tied at that tier; a random "Like" adds +50/+20 mental to every human, so keep rival families' gates more than 50 points away). The quest-gated skills that bite in practice (full 15-skill register in doc 22 §3.1): the Lazarus Pod (T12; the only locator-flag gate in the tree -- it unlocks after the investigation leg of the first seed mission, mid-quest, so it cannot be pre-bought), the Skimmer (T9; a later mission, which one is open), Night Sight (T9; "One Before Me"), Abyss Glow (T20; "Mother"), TideRipper Twin Jet (T19; "Still Listening"), Deep Scan-500 (T23; "Signal Chain"), the Telescope ("The Warehouse"), the Gyro Deck (T13; "A Place to Land"), the Farming/Elysium Bed (T19, 1 SP; "Gift from Above", the quest after the first launch), and the Recovery Grinder (T23, 4 SP; gate unnamed). 43+ named quests exist at 0.8.0; only the opening chain is listed spoiler-free in §5. Committee quests (U03+) gate on real-world elapsed time -- "it hasn't triggered yet" is often literally a matter of waiting a day or two; reload/reboot are the standing workarounds for known trigger bugs (HF2 fixed two quest-blockers; if an objective looks stuck, reload before assuming it's broken).

Run length (launch-era community): 20-35 h casual to "done for now"; 40-60+ thorough; sandbox continues after story. EA content currently ends mid-story (expectation-setting: there is no ending yet), then open play.

When Robert asks "what should I do next": 1. Answer from mechanics + preparation first (charge economy, range, storage, skills). 2. Route via doc 13 (strategy) and doc 11's early sections; doc 17 (quest compendium) only when he's AT the quest and wants specifics -- and offer the spoiler gate before objective details. 3. Never pre-reveal what a location contains, who speaks, or why. The map-reveal chain in doc 12/17 is mild-spoiler routing -- use its shape ("powering that beacon opens the next region") without its contents. 4. Log his choices/progress in praxium (decisions/scratch), never in docs/.

11. Interface essentials#

CONFIRMED binds (0.8.0): U skills · I enhancers · B build · C melee · E reload / interact · G inspect · Q cancel (hold: reel a cable in) · F switch ammo / connect cable to equipped weapon / flip socket mode · T hotbar quick-bind/re-equip · Page Down container access-control · RMB item context menu · Space/Ctrl gyro throttle [the first eight from Robert's own screenshots]. REPORTED launch-era (probably still right): WASD boat, Space anchor, Tab player menu, 1-8 hotbar, J clear markers. Grab/cable interactions are contextual across E/F/V and partly hardcoded. Skill tooltips carry an "Added in X.X.X" version stamp -- the UI itself dates content.

Options: 7 tabs (Video, Graphics, Audio, Game, Interface, Mouse, Gamepad). 16 HUD toggles ship, including full HUD-off. Accessibility: colorblind modes, subtitles (on by default), UI scale, hold-vs-toggle sprint/crouch, head-bob removal (U05), per-axis inverts. Controller support "full" per Steam; Deck = Playable (text legibility is the one failing item; U05-HF1 improved handheld UI scale after that rating).

Boat cab = ten dashboard monitors (speed, throttle, compass, coordinates, diesel, electricity, hull %, weather incl. wave height + wind, time). No depth sounder. Remember the dashboard/scanner rule (§5.6). PECO terminals are the game's computer surface: Data Logs, Space, Committee, Quests, Survival Guide, Cameras + a cartridge slot (playable retro minigames + Amiga-tracker music cassettes -- the loose peco-engine.js we found in the install is their engine; the minigame cartridges double as human-growth inputs). The PECO Grid Scan 2000 module renders your grid as a zoomable node map (per-node watts, fills, on/off state). The rocket's MOSES panel runs a sequential checklist (STARTUP -> PROPELLANT -> POWER -> GROUND -> PAYLOAD -> WEATHER -> SYSTEM CHECK -> LAUNCH) with a T-60 s countdown and ABORT.

Performance levers if needed: shadow quality is dominant (sg.ShadowQuality=0); known heavy scenes around the Lazarus Complex; 32 GB RAM recommended is real (HF2 fixed a critical memory issue). Crash lore in doc 15 §crash-table. Beta-era: the 5.8.1 engine's ray-tracing shaders crash on 2024-vintage NVIDIA drivers (nvgpucomp64.dll 0xc0000409) -- update the driver, or launch with -noraytracing.

12. Adjudications & open contradictions (cross-doc, settled here)#

13. Corpus health (what to trust, what's thin)#

Verification: all 20 swarm docs fact-checked in place (12-31 claims each, ~230 corrections; doc 08's verifier died on a transient API error mid-swarm and was re-run separately -- 17/18 clusters verified exactly, one aggregator-derived cable recipe corrected). The verifiers' systemic finding: the corpus's defect classes were real but bounded -- roadmap-as-shipped, launch-era-as-current, community-numbers-as-official, a handful of fabricated names -- all corrected, residuals fenced as UNVERIFIED in each doc. Since then: docs 21-22 (builds, skill tree) added 2026-08-24; fold-in batches 2026-08-27 (8 first-hand items) and 2026-09-01 (the mega batch: ~20 first-hand items + the flow-research reports folded across docs 05-22; doc 10 rebuilt as v2.0.0). Then 2026-09-03: the Eden Cradle / Black Vein companion batch (Stainless Steel's source, the tower intake trap, Angel TTK + the ArchAngel, the Black Vein hazard profile and site, Eden Cradle's two grids, NDNS huts, the ore district; docs 06 v1.4.0 / 09 v1.4.0 / 10 v2.1.0 / 12 v1.2.0 / 17 v1.2.0 / gas.md v1.0.2). Then 2026-09-05: the rack-screen batch -- two calibration screens reproduce the mined growth tables to the point, the requirement operator is meet-or-exceed on screen, the 100 kg rack hold and item masses, load-scaled growth time, backpack feeding, samples via the Replica Scanner and Camera Drone, the Seats of Care / The Forbidden Fruit objectives from the mine, Black Vein Station as built, the slime colour drift (docs 06 v1.4.1 / 09 v1.4.1 / 10 v2.2.0 / 12 v1.2.1 / 17 v1.3.0 / 18 v1.2.0 / 23 v1.1.0; the Council Cookbook page's watch-outs now read "do not add ... also qualifies as").

Systemic blind spots (all from source 403s, not laziness): SteamDB (no buildid history), Nexus Mods (mod scene is snippet-level -- ~167 mods reported), Reddit (only reachable when Robert pastes a thread -- which is how the Typhoon two-port question was settled), PCGamingWiki. The U05 endgame (Project Eden, committees' mechanics, quests 36-43) is thin everywhere -- the wiki lags the live build by one update; five U05 quests are wiki redlinks. Robert's playthrough is now the first-hand documentation for the growth and launch era.

The wiki (thelastcaretaker.wiki.gg) is canonical and alive (594 articles, edits within days of content) but broad-and-shallow: 52% stubs, a database not a guidebook, no controls page, no power/cables mechanics page, one patch-notes page. Its Cargo API does NOT exist (action=cargoquery returns an internal MWException -- probed 2026-08-24 by the doc-21 researcher); the machine-readable paths are per-page ?action=raw and the core MediaWiki revisions API. Struck as sources: the aggregator sites last-caretaker.com and thelastcaretakergame.org (stale, interpolated, falsely attributed -- electricity report §5.13).

Datamine (data/datamine/): pass 1 (2026-08-25) mined the skill tree, XP curve and point economy EXACT (no usmap needed -- retoc + exe-string-pool method; re-run recipe in the REPORT); pass 1.5 (2026-08-29) decoded the module assets by tooltip calibration (Large Battery 18 kWh, tank capacities, the solar ladder, purifier row); pass 2 (2026-09-01) decoded every module data asset byte-exact without a usmap (the property map's FName keys are in the package name table; 27 of 31 calibration anchors hit; parsed/modules-mined.json + tools/modules_parse.py): Large generator 33 kW / 66,000 L/h CO2, Sabatier 60,000 L/h + 5,000 L/h fresh water, purifier rates in L/h, Distillation Tower 20,000 L/h oil -> 10,000 diesel + 3,600 petrol, stabilizer 8 kW, Recycler/Fabricator standby draws, ExtraLarge petrol tank 10,000 L already in 0.8.0 data, plus crafting recipes for 70 assets (doc 09 fold pending). Still open: enhancer ladders and the Lazarus Rack's draw (both Blueprint Kismet), engine cylinders (outside the module set), barrels and pumps (no resource data), the save GVAS property tree beyond the naive scan (shadow-install §6.1). Flow-research fleet (2026-08-29): five reports under docs/ref/flow-research/ (electricity, fuel, water, gas, transfer-layer) are the numbers layer for everything that moves through a cable; doc 10 v2.0.0 is their synthesis. Growth-research fleet (2026-09-01): three reports under docs/ref/growth-research/ (seed-logistics, quality-and-committees, food-chain) plus datamine pass 3 (humans-mined.json: 41 professions, 15 foods, 59 memory items, mutations, likes, capsule reward tiers, 32 loot tables, byte-exact on the 5.5 build) are the human-growth layer; doc 23 is their synthesis and doc 18's tables are superseded where doc 23 §2.3 says so.

Mod track (2026-09-05..07): docs/mods/ is a second corpus -- all 176 Nexus mods downloaded, fingerprinted and classified (00-mod-library.md), five technique cores and twelve domain guides (01-17, each with a starter template under templates/mods/), the unified MODDING-FRAMEWORK.md v1.0.0 (decision tree, pipeline, methodology; reviewed 2026-09-07) and the house toolchain tools/mods/ (the tlcmod library, validated by rebuilding the shipping mods byte-identically). What is actually mounted on the Shadow, with its crash bisect and rollback, is 18-installed-mod-stack.md. Trust order inside that corpus: file evidence ([file]) > first-hand in-game reads > [catalog-only] Nexus text.

Refresh triggers: any new buildid on the Shadow -> re-run inventory + re-anchor docs 01/03 and the mod post-update drill (doc 18 §7: every pak is suspect until re-verified); Update 06 announcement -> doc 03 + roadmap sections. 5.5 stable release -- FIRED 2026-08-31 (buildid 25056839 on the Shadow): the next-patch batch is pending Robert's go -- shadow-install bump, full datamine re-run (retoc 0.1.5 with --version UE5_7 reads the 5.5 container; the beta exe identifier pool is already regenerated, closing the pass-2 blocker), verify the 11 new containers + Bulk Battery against the mined dormant content, re-anchor docs 01/03, lift the beta-era caveats. First death, first seed run and first launch -> settled (§12).

14. Doc Index (the routing table)#

Companion tools worth handing Robert when relevant: the1killer.github.io/LastCaretakerMap (open-source interactive map -- best in class), thelastcaretaker-tools.com (human growth calc + save import), turvarya's Steam guides (human growth), Caldason's range guide (oil doctrine -- pre-U05 numbers, doctrine now inverted).

---#

Provenance labels (added v1.1.0).

Evidence hierarchy, strongest first: player purchase or stat-panel read > in-game tooltip > screenshot digit read > datamine or save > wiki > guide.

v1.1.0 pass (2026-09-01) -- what changed. Three things, in order of how much they move the document. (1) The point economy is no longer estimated. DT_VoyageCharacterLevels was mined on 2026-08-25: leveling to the cap grants exactly 700 enhancer points and 305 skill points, on published bands (§3.2). The v1.0.0 arithmetic built on TheGamer's launch-era "sets of five" and landed on ~170 from leveling -- that figure is retired, it was low by a factor of four. The all-ten-maxable finding §3 argued from player reports is now closed by game arithmetic instead of testimony. (2) Vitality's "C tier" is an early-game verdict and the combat era has arrived. With a rifle in hand, enemies that burst you, and no armour system in the game at all confirmed 2026-08-28, the right call is now 15-20 points of Vitality bought from the float rather than zero (§2.3). Field Medic and Shock Absorber still skip. (3) Power Management may not do in the world what it does on the panel. Robert's panel reads exactly -225 W at 20 levels -- the formula is confirmed -- but three independent players report idle draw stuck near 260 W regardless of investment. That is a display-vs-measured gap and it is now a live bounty (§2.4), not a settled stat. Re-confirmed in this pass (2026-08-24): the live wiki Enhancers page still lists exactly ten enhancers with the per-level values the corpus already carries (+1 Vitality, +1 kg Haul Mastery, 1%/lvl Improved Hydraulics, 0.5%/lvl Engineering, +1 m Pressure to a 110 m total, 0.5%/lvl Scrapper, +20 Wh Battery to 2,500 Wh, 0.5%/lvl Power Management 250 W -> 125 W, +2.5%/lvl Field Medic, -0.5%/lvl Shock Absorber) and states "a maximum of 100 levels can be invested into every enhancer", with points "acquired through leveling and quests". Two things changed versus the corpus, both documented in full below:

1. The "you can only max about five of ten" scarcity premise is obsolete. Three independent post-Update-2 player reports say the opposite -- the newest, dated 2026-06-25 (one day after Update 05), reports 395 enhancer points left over after maxing enhancers. confirmed by three sources
2. The "Power Management is broken" warning is resolved. Patch 0.6.1.606604 (2026-01-12) states verbatim: "Fixed the power optimizer enhancer not working as intended." That is five months before this anchor. confirmed -- patch note

1. Orientation: what this document is for#

The Last Caretaker gives you two separate currencies at level-up. Skill points unlock recipes on a 25-tier tech tree; Enhancer points upgrade the ten permanent stats on your chassis. This document is only about the second one. It answers a single question in as much detail as the sources allow: given the points in front of me right now, at this level, what do I buy?

Three facts frame every decision:

That second row is the most important correction in this document, and it changes what a "build" means here. Read §3 before you read any build.

A note on what is and is not community-sourced. This game's enhancer meta is thinly documented. There is no canonical community "build guide" with named archetypes. What genuinely exists is a Steam thread in which seven named players posted their actual point distributions, plus a priority-order consensus thread. Builds below are labelled either community (traced to a real post, cited, with the poster's name and date) or synthesized from verified mechanics (mine, derived only from the corpus-verified numbers). Nothing here is attributed to "the community" without a citation.

2. The enhancer roster#

All per-level values and caps below are confirmed -- wiki Enhancers, re-verified 2026-08-24, and already fact-checked into docs/05-caretaker-systems-survival.md. The verdict column is my S/A/B/C tiering of the community consensus described in §2.1, not an official ranking.

2.1 Where the verdicts come from#

Two threads carry essentially all of the published priority reasoning.

Note the disagreement on Vitality: TheGamer ranks it third; every actual player distribution in the Steam thread ran it at 0-30. The players are operating on far more play time, and the corpus already establishes why they are right -- Vitality does not regenerate in the field, so the answer to "I am low" is retreat, which extra maximum Vitality does not help with. I follow the players.

2.2 Two build-relevant bug questions, both now resolved#

The corpus (doc 13, failure pattern #11) flags "points sunk into a non-functional stat" as an open risk, citing 0.6.x-era reports that Power Management and Scrapper did not work. Both are now settled:

You no longer need the corpus's "spend 10 points and watch the W readout before committing" hedge for Power Management. It is worth doing once anyway, because it costs nothing and this is Early Access. v1.1.0 correction: do it. §2.5 explains why that hedge came back.

2.3 The Vitality verdict, revised for the combat era [v1.1.0]#

⚠️ This supersedes the "Vitality is C-tier, run it at 0-30" consensus for any player past the early game. The old verdict is not wrong about why -- it is wrong about when.

What the launch-era consensus assumed. Every distribution in §2.1 was posted Nov-Dec 2025 by players who (a) were avoiding combat because early weapons are weak, (b) could always retreat, and (c) had scarce points. Under those three conditions Vitality genuinely is the worst buy: it does not regenerate in the field, so more maximum Vitality does not change the correct response to being hurt, which is to leave.

What changed.

The revised guidance: 15-20 points of Vitality, bought from the float once you are fighting rather than avoiding. That is a deliberately small number -- it is a margin, not a build -- and it comes out of surplus, never ahead of Battery, Power Management or Haul Mastery. Field Medic and Shock Absorber still skip: Field Medic speeds recovery you can also get by standing at a Heal Station, and Shock Absorber solves a problem ladders and a charged battery already solve for free.

Provenance: the era change is first-hand 2026-08-28; the no-armour finding is confirmed absence plus [datamined] never-gated armour assets; the 15-20 figure is a companion recommendation, not a community consensus and not a measured optimum.

2.4 Three formulas confirmed exact, first-hand [v1.1.0]#

The wiki's per-level values were reported until Robert read them off his own stat panel. Three are now first-hand 2026-08-25, build 0.8.0.612251 and exact -- the panel arithmetic pins both the per-level step and the base:

Two of those bases (500 Wh, 100 kg) were already confirmed elsewhere; the value here is that the curves are now first-hand rather than inherited, which means the 100-level endpoints in §2 -- 2,500 Wh, 125 W, 200 kg -- are arithmetic, not testimony.

Mobility % and Extra Weight % are HUD stat-panel readouts, not enhancer fields: they are the live consequences of your current load against your current carry cap, which is why over-weight play shows up in both at once.

⚠️ The Power Management row is a DISPLAY confirmation only. The panel says 225 W. Whether the Caretaker actually draws 225 W is a separate question -- see §2.5.

2.5 🐛 The Power Management measured-draw bounty [v1.1.0, OPEN]#

§2.2 closed "is Power Management broken?" on a patch note. That was the right call for the patch-note question and the wrong scope for the behaviour question, which is still open.

The reports (docs/ref/flow-research/electricity.md §5.14): three players, independently, describe idle draw that does not move with investment --

"-50% power savings, but my power usage is STILL at around 260." -- WillieSea, corroborated by two others; the same poster separately reports "my power consumption is around 260 standing still", which is the un-enhanced 250 W baseline plus noise.

Why this is not settled by §2.4. Robert's panel reads 225 W at 20 levels, which proves the stat display applies the formula. It does not prove the simulation consumes at that rate. Those are different subsystems and the community reports are specifically about the second one.

🧪 The measurement that settles it, and it is cheap: stand still, unencumbered, torch OFF, and read the actual draw on the boat's electricity dashboard (or any grid gauge you are tethered to) before investing a block of Power Management points and again after. If the measured idle draw falls by the predicted percentage, the reports are stale; if it sits near 260 W either way, up to 100 enhancer points are being spent on nothing and every build in §4 needs re-ordering.

Until that read exists, treat Power Management as very likely working (the patch note is explicit and the display is exact) but not measured -- and do not spend past ~40 points in one block without checking the gauge.

3. The point economy -- doing the arithmetic explicitly#

This section supersedes the corpus's "~5 of 10 maxable per playthrough" figure. That figure is real, but it is a launch-era artifact, and the mechanism behind its obsolescence is documented.

3.1 What it costs to max#

⚠️ There are eleven enhancers, not ten [first-hand 2026-08-26; see 05-caretaker-systems-survival.md]. The undocumented eleventh is Equipment Handling ("Improves equipment handling 0.5%. This affects holstering, melee swing and reload durations") -- it is in no wiki page and no guide, but it is in the live panel and in the datamine's roster of eleven DA_Enhancer_* assets. This document keeps its arithmetic on ten because every published figure, every player distribution and the wiki roster do, and because Equipment Handling's 100-level cap is assumed rather than confirmed. Read "1,000 to max everything" as "1,000 to max everything anyone has written about". §6 records the correction against v1.0.0's rejection of an 11th category.

3.2 Where points come from#

Source A -- level-ups. DATAMINED EXACT [2026-08-25], and it is four times what v1.0.0 estimated.

DT_VoyageCharacterLevels (mined from buildid 23962331; data/datamine/parsed/xp-curve.json) publishes the per-level grant table outright. There is no estimation left to do:

Cumulative checkpoints, for planning: 45 enhancer points banked by level 5 · 125 by level 10 · 325 by level 20 · 700 at the cap. Skill points run 15 / 55 / 155 / 305 on the same checkpoints, against a 267 SP tree (22-skill-tree-progression-paths.md), which is why the tree finishes from leveling alone.

🔴 ~~Points per level: 5 reported, TheGamer 2025-11-10; total from levelling ~170 inferred~~ -- SUPERSEDED and retired. TheGamer's "awarded in sets of five" described the launch-era, eight-enhancer game and was low by roughly a factor of four. jonnydoh's "5-10 per level" was closer in spirit (the grant does scale with level) but still an order of magnitude off in value. Every §3 figure below is rebuilt on the mined table.

Source B -- quests. This is the dominant source, and it is the mechanism that broke the old scarcity math. Summing the wiki-sourced Enhancer Point rewards already catalogued in docs/11-progression-missions-objectives.md:

The undocumented twelve are exactly the Update 05 endgame (Project Eden, Seats of Care, True Choices, Omega Nursery and friends). Late main-chain quests award 30-50 apiece, so this bucket is large but unquantified.

3.3 Floor, ceiling, and the gap#

The documented floor -- what you can prove you will get:

🎯 That closes the question the whole section was built to answer. The floor is now above the 1,000 points needed to max all ten enhancers, on documented sources alone -- before the twelve undocumented Update 05 quests contribute anything. All ten maxable is no longer an inference from three player reports; it is arithmetic. (It also clears 1,100, the cost of maxing eleven including Equipment Handling, with 115 to spare.)

~~Documented floor ~685~~ -- superseded; that figure carried the retired ~170 leveling estimate.

The Nexus corroboration is now an exact hit, not an approximation. The mod "Enough Enhancement and-or Skill Points to Max Skills at Level Cap" describes vanilla as granting 700 enhancement points and 305 skill points [reported -- modder's description via search snippet; the page itself is HTTP 403]. Both numbers match the mined table exactly. The modder was reading the same data table; v1.0.0 treated their figure as weak corroboration of an estimate, and it turns out to have been the true value all along.

The observed ceiling -- what players actually report ending with:

Grimble Grumble adds that this surplus has held "all the way back to update 2", and their post is dated one day after Update 05 shipped -- the closest thing to a current-build datapoint that exists.

The reconciliation. Documented floor 1,215; observed ceiling ~1,395. The remaining ~180-point gap is comfortably accounted for by the twelve undocumented Update 05 quests alone (late main-chain quests award 30-50 apiece). The upward scaling that v1.0.0 hypothesised in the per-level grant is now confirmed and quantified -- 10 -> 15 -> 20 -> 25 by band -- and is already inside the 700. What remains a range is only the quest contribution above the documented 515.

3.4 Why the old "five of ten" figure was true and no longer is#

⚠️ CORPUS CONTRADICTION, FLAGGED. docs/05 §7.1, docs/11 §5.3 and docs/13 §5 all carry the "~5 of 10 maxable" / "1,000 points is far more than a single playthrough grants" framing. On the evidence above that framing is stale. Note that the corpus already contained the counter-evidence and hedged it -- doc 13 quotes Shane's "their goal seems to be to allow you to max them all" and calls the five-stat figure "now conservative". This document resolves that tension in favour of the surplus reading, on three independent post-Update-2 reports plus the patch history of the mechanism. Those docs should be updated on their next version bump.

3.5 What this actually means for your build#

It does not mean spending is unimportant. It means the shape of the problem is different from what the old guides describe:

1. You will not be permanently locked out of a stat. Finish the content and you can have everything. Nobody needs to agonise over "wasting" points on Shock Absorber at level 30.
2. You will be brutally short early. You have ~10 points when you first see the menu and roughly 60 by level 5. That is where builds are won or lost.
3. Therefore: a build is a spending order, not a spending allocation. The tables below are written as order-of-purchase, not as end-state pie charts.
4. The corollary is that under-spending is the real mistake. Points sitting unspent do nothing. The community's most-repeated beginner tip -- check Tab, unspent points show as green numbers -- matters more than picking the theoretically optimal stat.

3.6 Approximate points available by bracket#

Rebuilt on the mined grant table [v1.1.0]. The leveling column is now exact; the quest column is still [wiki]-sourced and assigned to the bracket where those quests typically fall, so the combined figure keeps real slop. Quest pacing varies enormously between players.

⚠️ The build scripts in §4 were budgeted against the old, much smaller bracket totals (~60 / ~85 / ~130 / ~410). They are therefore conservative, not wrong: every script now finishes earlier than written, and the surplus arrives sooner. Read the scripts as orderings -- which is what §3.5 already says they are -- and spend the extra as it lands. The one place the new numbers change advice rather than pace is §2.3: there is room for the Vitality margin without touching the S-tier core.

4. The builds#

Six builds. Each gives you a philosophy, a priority table, a level-bracket spending script, and an explicit list of what it skips. Every script is budgeted against the §3.6 bracket totals, so the numbers add up to something you will plausibly have.

Read the Grid-Runner first even if you intend to play something else -- it is the baseline the others deviate from.

4.1 The Grid-Runner -- generalized all-rounder#

• Kind: generalized
• Source: synthesized from verified mechanics
• Tagline: Never be the Caretaker who walks home flat.

Philosophy. This build treats the internal battery as the game's real difficulty setting and buys operating radius before anything else. Battery and Power Management are bought together deliberately: Battery multiplies how much charge you hold, Power Management multiplies how long that charge lasts, and stacking both compounds -- at 100/100 in each you go from 500 Wh at 250 W (2.0 h idle) to 2,500 Wh at 125 W (20.0 h idle, a 10x endurance swing on the corpus's own arithmetic). Haul Mastery follows because encumbrance is a power problem, not just a speed problem. Pressure is bought late but not skipped, because it is the only stat that gates content.

Priority table:

Level-bracket spending script:

Deliberately skips:

4.2 The Salvage Baron -- salvage / industry specialist#

• Kind: specialized
• Source: community -- Purple Legacy, Steam "How have you distributed your enhancement points?", 2025-11-20 (Haul Mastery 100 / Battery 100 / Scrapper 100 / Power Management ~75 / Hydraulics ~50 / Vitality 0 / Pressure 0), with quaixor (2025-12-04) independently proposing the same Haul 100 / Scrapper 100 / Battery 100 core.
• Tagline: Three hundred kilos of scrap and the power to walk it home.

Philosophy. Two named players arrived at the same triangle independently: Haul Mastery, Battery and Scrapper all to 100, with Pressure at zero. The logic is that the salvage loop -- strip, haul, recycle, fabricate -- is the entire economy, and each of those three stats multiplies a different stage of it. Purple Legacy explicitly traded away underwater content to fund it. This is the strongest build in the game for the first twenty hours and the weakest for content access.

Priority table:

Level-bracket spending script:

Deliberately skips:

4.3 The Abyss Diver -- deep-dive specialist#

• Kind: specialized
• Source: community -- "Sad Cat Is In San Francisco!", Steam distribution thread, 2025-11-22 (Pressure 100 / Haul Mastery 60 / Hydraulics 30 / Scrapper 10 / Battery 10 / Vitality 0 / Engineering 0 / Power Management 0, at level 25 with 240 points still spare), plus Raytenecks (2025-11-20, Pressure 70 first) and Ed130-The Vanguard's gating rule in the "best points spend?" thread.
• Tagline: The only stat that opens doors.

Philosophy. Pressure is the one enhancer that is not an efficiency multiplier -- it is a key. Content sits behind depth, and no amount of Battery substitutes for the ability to be at 90 m without taking tick damage. Ed130-The Vanguard's rule is the practical one: ~40 points clears the main questline's diving, and you max it only if you intend to visit other underwater locations. Sad Cat maxed it, reported the deep sites "a pain to get to", and still had 240 points spare -- which is itself the best evidence for §3's surplus finding.

Battery matters here for a second-order reason the corpus documents: diving below your limit increases battery drain on top of the Vitality ticks, so an over-pressure dive burns charge faster exactly when you can least afford to swim home.

Priority table:

Level-bracket spending script:

Deliberately skips:

⚠️ Depth reality check. Pressure caps at 110 m, but the corpus documents caves with entrances from 60 m to 196 m, and some quest dive sites sit deeper than a fully-upgraded Pressure stat allows -- they are designed to be done partly over-pressure. Maxing Pressure does not make you immune to the deep; it moves the line. Bring the tethered-cable technique regardless.

4.4 The Efficiency Core -- minimal-power specialist#

• Kind: specialized
• Source: synthesized from verified mechanics
• Tagline: Don't carry a bigger tank. Burn less.

Philosophy. Every other build buys capacity. This one buys multipliers and leans on infrastructure instead: Power Management, Engineering, Scrapper and Improved Hydraulics are all percentage reductions, and they stack against a base you deliberately keep small. The thesis is that the corpus's own charge doctrine -- stay tethered, charge off a storage battery on the grid, build forward recyclers -- makes raw Battery capacity partly redundant, and that points are better spent making every watt-hour do more work. Power Management alone takes idle draw from 250 W to 125 W, doubling the endurance of any battery size, including the stock 500 Wh one.

This is the highest-skill build in the document and the one most punished by sloppy play. It is also the most satisfying if you enjoy logistics, and it front-loads the stat the launch-era guides named first.

Priority table:

Level-bracket spending script:

Deliberately skips:

🪤 Gotcha. This build was actively dangerous before 2026-01-12, because Power Management genuinely did not work. It is safe on 0.8.0.612251. If you are ever playing an older build, do not run this.

4.5 The Expedition Chassis -- combat / long-range exploration#

• Kind: specialized
• Source: synthesized from verified mechanics
• Tagline: For the player who intends to be somewhere dangerous, on purpose, for a long time.

Philosophy. The consensus dismisses Vitality and Field Medic because combat is avoidable and healing is available at Heal Stations. That is correct if you avoid combat. Update 03 added Ram Sharks, Laser Sharks and ArchAngels, and Update 04 added deep vertical terrain -- open water stopped being neutral. This build is the deliberate contrarian answer: it buys the survivability pair that everyone else skips, and pairs it with Battery because a long expedition is a power problem before it is a combat problem.

Field Medic is the sleeper here. It governs both channels -- charge transfer 200 kW -> ~700 kW and healing 10 HP/s -> ~45 HP/s. On a build that expects to take damage and top up in the field, it is the only stat that shortens both recovery loops at once.

Priority table:

Level-bracket spending script:

Deliberately skips:

⚠️ Honest labelling. No player in any thread found ran this build. It is synthesized, and it runs against the documented consensus on Vitality and Field Medic. It is included because the consensus was formed in the launch-era 0.5-0.6 game, before three updates of new ocean threats, and because the corpus's own Field Medic numbers are better than its reputation. Treat it as a considered contrarian option, not a proven one.

✅ v1.1.0 -- half-vindicated. §2.3 revises Vitality upward for the combat era on first-hand evidence, and confirms there is no armour system, which is this build's core argument. What §2.3 does not endorse is the scale: it recommends 15-20 points from the float, not the 100 this build buys. Read the Expedition Chassis as the right instinct at four to six times the necessary dose. Field Medic at 100 remains unendorsed -- a Heal Station is free.

4.6 The Quest-Pace Balanced -- a real player's spread#

• Kind: community
• Source: community -- jonnydoh, Steam "How have you distributed your enhancement points?", 2025-12-04. Verbatim distribution: Vitality 30 / Haul Mastery 50 / Improved Hydraulics 60 / Engineering 20 / Pressure 100 / Scrapper 70 / Battery 50 / Power Management 70 = 450 points, describing "5-10 points per level" and noting they "forgot to spend accumulated points and suddenly maxed pressure."
• Tagline: What an actual playthrough looks like, unoptimised and fine.

Philosophy. This is included as a reality check against the tidy builds above. jonnydoh spread points across eight of ten stats, maxed only Pressure, and finished the content. It is proof that the game does not punish a generalist -- and its most instructive detail is the admission of forgetting to spend points at all. Under-spending is a bigger real-world cost than mis-spending. If you find the optimisation stressful, this is the build: keep everything moving, max Pressure when the dive gate appears, and check Tab.

Priority table:

Level-bracket spending script (reconstructed to reach the posted distribution, then extended):

Deliberately skips:

4.7 Build comparison at a glance#

End-state targets by build, showing where each one commits. Numbers are the Phase-4 targets from each script.

The row that matters: Battery, Power Management and Haul Mastery reach 100 in all six builds. That is the real consensus. Everything else is flavour.

⚠️ One row to read with §2.3 open: the Vitality row shows 0-100 depending on build, all of it set by the launch-era consensus. For any player past the early game the floor in that row should now be 15-20, taken from surplus. It does not change any build's identity; it changes what you do with points you were going to have spare anyway.

4.8 Optional flavour: Improved Hydraulics as a mobility lane [v1.1.0]#

🏷️ Play-style option, NOT consensus, NOT measured. Recorded because it was recommended in play (companion recommendation, 2026-08-28) and it is a reasonable reading of the stat -- but no community source proposes it and no first-hand measurement supports it. Do not treat it as a sixth verdict tier.

The reading. Improved Hydraulics is documented as lifting capacity and lifting power (+1% / -1% per level; the live panel words it "+1% lifting capacity and +1% efficiency"). Every build above therefore scores it as crate work -- base logistics, skippable if you do not move heavy objects. The alternative reading is that on a chassis that sprints a lot, the same efficiency channel is quietly paying rent all the time, because encumbrance in this game is a power tax before it is a speed penalty (over-weight idle draw reported at 400 W+ against a 250 W baseline, §2). A player who is habitually near their carry cap and moving fast is paying that tax constantly.

How to run it if you want to: treat Improved Hydraulics as a third efficiency stat behind Power Management and Scrapper, target ~50 by the mid-game and 100 from surplus, and pair it with Haul Mastery rather than instead of it -- Haul raises the threshold, Hydraulics lowers the cost of working near it.

Why it is fenced: the -1%/level power channel is documented for lifting, and whether it touches general movement draw at all is UNVERIFIED (§6 already flags that its -100% endpoint is unconfirmed). If the channel is lifting-only, this lane buys nothing a sprinter can feel. 🧪 The settling measurement is the same gauge read as §2.5: watch idle and moving draw at fixed load, before and after a block of Hydraulics points.

5. Switching lanes mid-game#

There is no respec -- confirmed across multiple threads, and the community has asked repeatedly. Points spent are spent. But because the lifetime supply now exceeds the 1,000-point cost of everything (§3), "switching lanes" here does not mean recovering points. It means re-sequencing what you buy next, and bridging the gap with gear while the points accumulate.

5.1 The four common lane changes#

5.2 The gear bridge#

Every enhancer weakness has an equipment answer that costs skill points instead of enhancer points. Use these while the points accumulate:

5.3 The one rule that beats every build#

Check Tab. Unspent points show as green numbers. jonnydoh -- a real player who finished the content -- lost track and accidentally banked enough to max Pressure in one go. Every thread repeats this tip. A mediocre allocation that is actually spent beats an optimal one sitting in the menu.

6. Gaps & Unverified#

Numbers that could not be established:

• ✅ ~~Exact enhancer points per level on 0.8.0.~~ -- CLOSED [datamined 2026-08-25]. DT_VoyageCharacterLevels publishes the table: 10/level at L2-4, 15 at L5-9, 20 at L10-20, 25 at L21-35 = 700 to the cap, alongside 305 skill points. TheGamer's "sets of five" was low by ~4x and is retired; jonnydoh's "5-10 per level" was directionally right about scaling and wrong about magnitude. §3.2 is rebuilt on the mined table.
• Enhancer point rewards for quests 29-32 and 36-43 (12 quests, the Update 05 endgame). No wiki pages exist. This is the single largest hole in the point economy and the reason §3.3 reports a range rather than a total.
• Total enhancer points available on a 100% playthrough. Half-closed. The leveling half is exact (700). The quest half is documented to 515 across 27 quests with wiki pages, with twelve Update 05 quests unpaged and unquantified -- that residual is the entire remaining uncertainty, and it is what puts the true total somewhere between the 1,215 documented floor and the ~1,395 players report.
• ✅ ~~Whether the level-up grant changed when the cap went 30 -> 35~~ -- moot for planning: the mined table is the current build's table and covers levels 1-35, with the 25/level band starting at 21. Whether it differed pre-U05 is now a history question, not a build question.
• Base Vitality (100) and base dive depth (10 m) remain inferred from the enhancer math, per the corpus. Not independently confirmed here. Contrast §2.4: the Battery (500 Wh), Power Management (250 W) and Haul Mastery (100 kg) bases are no longer inferred -- they are first-hand.
• 🐛 Whether Power Management reduces MEASURED draw, not just the displayed figure. Three independent players report idle draw stuck near 260 W regardless of investment. Robert's panel confirms the display formula exactly (-225 W at 20 levels) but nobody has read a gauge before and after investing. OPEN -- see §2.5. This is the single highest-value cheap measurement in the document.
• Field Medic's 200 kW -> 700 kW transfer figures carry the wiki's own "verification needed" flag. They are corroborated by an independent player report of "+250% recharge rate at 100 points" (the same 3.5x), which is strong, but neither source is a measurement. REPORTED, not confirmed.
• Whether Improved Hydraulics' power channel touches anything beyond lifting -- the basis of the optional lane in §4.8. UNVERIFIED, and the -100% endpoint is separately unconfirmed (below).
• Whether Improved Hydraulics' -1%/level power reduction actually reaches -100% (i.e. free lifting) at 100/100, or is capped/floored somewhere. The wiki states the per-level value without describing the endpoint. Flagged in the corpus too; still UNVERIFIED.

Claims that contradict the corpus (flagged per protocol):

Claims found and rejected:

• 🔴 ~~An "11th enhancement category."~~ REJECTION SUPERSEDED [v1.1.0]. v1.0.0 rejected the Nexus mod's mention of an 11th category on the strength of the wiki's ten-entry Enhancers page. The modder was right and the wiki is incomplete. The eleventh is Equipment Handling -- "Improves equipment handling 0.5%. This affects holstering, melee swing and reload durations" -- confirmed in the live panel [first-hand 2026-08-26, see 05-caretaker-systems-survival.md], present in the datamine as the eleventh DA_Enhancer_* asset (DA_Enhancer_Equipment_Handling / BP_EnhancerInfoLogic_Equipment_Handling_C), and separately flagged in 19-interface-controls-settings.md. The mod's 1,100 total is exactly 11 x 100. This document's arithmetic still runs on ten (§3.1) because every published figure and every player distribution does, and because Equipment Handling's own 100-level cap is assumed, not confirmed. The lesson for the corpus: "the wiki lists ten" was never evidence that ten is all there are.
• "Every Enhancer Point should go into Power Management for the first 50 levels" and "even standing still burns 5W" -- surfaced in search snippets from last-caretaker.com / thelastcaretakergame.org. Rejected. Those domains are on the corpus's unreliable list, and the 5 W figure contradicts the wiki-confirmed 250 W base draw by a factor of 50. Recorded here so a future session recognises and rejects it again.
• "A minimum battery capacity is required to dismantle certain objects." Surfaced in a search summary, so I fetched the underlying thread to check. Not present -- the thread contains no such claim. Discarded as a search-summary artifact, not a game mechanic.

Sources that could not be used:

• nexusmods.com/thelastcaretaker/mods/16 -- HTTP 403, abandoned permanently. Its description was reachable only through search snippets, which is why the "700 vanilla points" figure is tagged reported and used only as corroboration of an independently-derived number, never as a primary source.
• thelastcaretaker.wiki.gg/api.php?action=cargoquery -- returns internal_api_error_MWException; this wiki has no Cargo tables. All wiki data was taken via ?action=raw instead. Future sessions can skip the Cargo attempt.
• Reddit -- not reachable by this agent's fetcher, per corpus precedent. No Reddit consensus is represented anywhere in this document.
• YouTube watch pages -- not fetched, per protocol. One relevant video title was found and is recorded in Sources as existence evidence only; none of its content is used.

Structural caveats:

• Most of the priority consensus is launch-era. The two threads carrying the reasoning ran Nov 2025 - Jan 2026, on a game with a level cap of 25-30, a broken Power Management stat, and no quest-awarded enhancer points. Their ordering has survived (nobody has argued against Battery-first since), but their scarcity framing has not.
• Every player distribution in §4 is from Nov-Dec 2025. They are real and citable, but none of them were built on 0.8.0. No 2026-era player has posted a full distribution anywhere reachable.
• The bracket totals in §3.6 are inferred, built by assigning wiki-documented quest rewards to the level bracket where those quests typically occur. Quest pacing varies enormously between players; treat them as planning aids with real slop.
• Three of six builds are synthesized, derived only from corpus-verified mechanics. They are labelled as such at the top of each build. The Expedition Chassis in particular runs against the documented consensus and says so.

Sources#

All accessed 2026-08-24.

First-hand ground truth (local):

• C:/last-caretaker/docs/ref/shadow-install.md v1.0.1 -- EA 0.8.0.612251, buildid 23962331, UE 5.7.4, save format decoded
• [datamined 2026-08-25] C:/last-caretaker/data/datamine/parsed/xp-curve.json -- DT_VoyageCharacterLevels from buildid 23962331: the exact per-level skill-point and enhancer-point grant table, 305 / 700 to the cap, 99,000 XP. Re-run recipe in data/datamine/REPORT.md
• [datamined 2026-08-25] .../parsed/enhancers-mined.json -- eleven DA_Enhancer_* assets (the eleventh being Equipment Handling) and enhancer_points_available: 700. Note its blocker: the per-level values live in Kismet bytecode, not asset data, so the mine confirms the roster and the budget but not the ladders
• first-hand 2026-08-25 in-game stat panel at a 20 / 20 / 10 spend: Battery 0.9 kWh, Power Management 225 W, Haul Mastery 110 kg -- the three exact-formula confirmations in §2.4
• first-hand 2026-08-26 in-game enhancer panel: the eleventh enhancer (Equipment Handling), Scrapper's four effects, Improved Hydraulics' wording -- recorded in 05-caretaker-systems-survival.md
• first-hand 2026-08-28 the combat era: rifle in hand, burst damage taken, no armour system available -- the basis of §2.3
• C:/last-caretaker/docs/ref/flow-research/electricity.md §2.6 and §5.14 -- the enhancer electrical economy, the encumbrance power tax (400 W+ over-weight), the Field Medic corroboration, and the three-player Power Management measured-draw reports behind §2.5
• C:/last-caretaker/docs/05-caretaker-systems-survival.md v1.1.0 -- the ten enhancers with verified per-level values, battery/vitality model, charge doctrine, fall-damage threshold
• C:/last-caretaker/docs/11-progression-missions-objectives.md v1.1.0 -- the per-quest Enhancer Point reward table that §3.2's 515-point sum is computed from, level cap 35, skill tiers
• C:/last-caretaker/docs/13-strategy-optimization.md v1.1.0 -- enhancer priority consensus, the broken-stat warnings this document resolves, gear bridges

Community wiki (thelastcaretaker.wiki.gg):

• https://thelastcaretaker.wiki.gg/wiki/Enhancers?action=raw -- all ten enhancers, per-level values, 100-level cap, Power Management 250 W -> 125 W table, "acquired through leveling and quests"
• https://thelastcaretaker.wiki.gg/wiki/Caretaker?action=raw -- XP sources, "each level grants the corresponding tier of Skills, along with Enhancer Points"; page is a stub and gives no per-level number
• https://thelastcaretaker.wiki.gg/api.php?action=cargoquery -- no Cargo tables (MWException); recorded so it is not retried

Patch notes:

• https://changelog.gg/games/the-last-caretaker-1783560/updates/2026-01-12-patch-0-6-1-606604-db45 -- Patch 0.6.1.606604, 2026-01-12: "Fixed the power optimizer enhancer not working as intended." The source that resolves the Power Management question.

Steam community discussions (each build's attribution traces to a specific post here):

• https://steamcommunity.com/app/1783560/discussions/0/681860392750808508/ -- "How have you distributed your enhancement points?" (opened 2025-11-20). The primary build source. Purple Legacy, Raytenecks, Sad Cat Is In San Francisco!, rditto48801, jonnydoh, quaixor, Xanman64 distributions; Sad Cat's "level 25, 240 spare points" datapoint.
• https://steamcommunity.com/app/1783560/discussions/0/690871777621988583/ -- "best points spend?" (2026-01-04). Priority consensus; Ed130-The Vanguard's "40 or so for the main questline" Pressure rule; Shane's "you get almost all of them anyway"; WillieSea's Power Management report.
• https://steamcommunity.com/app/1783560/discussions/0/669474225528382054/ -- "enhancement points more please" (2025-11-08). Playbahnosh: launch-era cap 25, "max out 2 maybe 3 enhancements", quests did not award enhancer points at launch. Back_Door_Gaming + DreadedOne (2026-04-02): "10 left over after filling every single one to 100" / "I have 175 extra now".
• https://steamcommunity.com/app/1783560/discussions/0/681859758979902918/ -- "Why no respec?" (2025-11-12, live through 2026-06-25). Grimble Grumble, 2026-06-25: "82 skill points and 395 Enhancer points left over"; "instead you'll receive more of both than you can even spend"; surplus true "all the way back to update 2." The most current datapoint in this document.
• https://steamcommunity.com/app/1783560/discussions/0/690870479611882270/ -- "Scrapper Not Working As Intended?" (2025-12-22). deltahazer: Scrapper is percentage-based and "working as advertised"; SmilesLikeJoker: Power Management was the actual bug.

Press / guides:

• https://www.thegamer.com/the-last-caretaker-best-enhancers-level-first/ (2025-11-10) -- "Points are awarded in sets of five as you level"; 100 max per category; eight enhancers (launch-era); Battery + Power Management first ordering
• https://steamcommunity.com/sharedfiles/filedetails/?id=3600430670 -- TechRaptor Starter Guide (2025-11-06), hosted on Steam: "you have 10 unused enhancer points; for now, I recommend putting 10 points into Power Management"

Referenced but not used as evidence:

• https://www.nexusmods.com/thelastcaretaker/mods/16 -- "Enough Enhancement and-or Skill Points to Max Skills at Level Cap". HTTP 403, abandoned. Description reached only via search snippet: vanilla 700 -> 1,000 points, skill points 305, and an unconfirmed "11th enhancement category" claim. Used only as weak corroboration of the §3.3 floor.
• https://www.youtube.com/watch?v=Wf-11SUMA2k -- "Don't Waste Your Skill Points - The Last Caretaker Build Guide" (Dec 2025). Title only -- watch page not fetched per protocol. Recorded as evidence that video build guides exist and are launch-era.

v2.1.0 GATE pass (2026-09-01). The mine established that fifteen tier-listed skills carry a custom_requirements object; play established which gates those objects are. Settled here: the Lazarus Pod's requirement is not the seed quest's completion but a locator flag set mid-quest -- by the underwater-investigation leg beneath the first seed vault (~45 m dive), before you lift a seed -- so the pod IS buyable on the first seed run [first-hand: purchase refused 2026-08-28 at character level 15, granted 2026-08-30]. Night Sight's mined requirement is validated first-hand (refused before One Before Me, available after). The CS-17 Courier Skimmer is mission-gated in practice as well as in the data (purchase blocked "until another mission"; which quest is an open bounty). Abyss Glow reads MOTHER and the TideRipper(TM) Twin Jet reads STILL LISTENING on their in-game tooltips [beta-era]. The Elysium Bed -- the wiki's "Farming Bed" -- is the Gift from Above reward, not the "possess a seed" item-unlock its wiki infobox describes. Section 3.1 is now the complete register. Datamined prices folded into section 2: PECO Grid Scan 2000 T15 / 2 SP (the corpus's T16 was wrong), the four wall sockets at 1 SP each (T15/16/17/18), CO2 Tank T16 / 3 SP, and v2.0.0's T9 relocations carried into the band tables. The version anchor does not move: everything is 0.8.0.612251 unless tagged [beta-era] (Update 5.5 "Deck Drop" public beta, 0.8.5.650920, buildid 24990438, UE 5.8.1, opted into 2026-08-29).

v2.0.0 DATAMINE pass (2026-08-25). The tree is now sourced from the game data itself (data/datamine/parsed/skilltree-mined.json: 200 VoyageSkill assets + DT_VoyageCharacterLevels, wiki-cross-validated): exact XP curve (200 XP at L1, +150/lvl to L25, +200/lvl to L35; 99,000 total to cap), exact point grants (skill: 3-9 rising then flat 10/level = 305 total; enhancer: 10/15/20/25 by band = 700 total + quest grants), 154 tier-unlocked skills costing 267 SP -- the tree completes from leveling alone with ~38 SP spare -- plus 44 quest/vault-gated skills. Settled for good: Petrol Engine Cylinder T9/1SP (the item page beat the nav) and CS-17 Courier Skimmer T9/3SP (the corpus-wide T13 was wrong); Bolt Pistol T5/2SP matches the player purchase receipt exactly. Machine truth: docs/data/skilltree.json, regenerated from the mine; re-run recipe in data/datamine/REPORT.md.

v1.1.0 first-hand correction pass (2026-08-25). A live in-game screenshot (level-5 player on the current build) + a re-verification sweep corrected the dataset: Bolt Pistol costs 2 SP -- confirmed by actual player purchase 2026-08-25 (an interim screenshot misread of "1" is retracted, and with it the "TheGamer proven wrong" downgrade: its sole-source prices return to [reported] -- uncorroborated, but with no proven error); Wall Step (T5/1), DONK Pallet (T5/1), DONK-Lift 3200 (T5/2) and Replica Scanner (T3/1) are now priced first-hand; LumiLite T2, DuraPack T3, Electric Pistol T3 settled; the T2 "blue container" tile is the Basic Electric Engine Cylinder. The in-game tooltips carry an "Added in X.X.X" version stamp the wiki does not mirror -- in-game reads and Steam patch notes are the only U05-membership sources. Documented tree cost: 207 SP across 122 priced unlocks; 18 still unpriced. The Petrol Engine Cylinder tier conflict (nav 13 vs item page 9) stands -- an in-game check at level 9 settles it. Machine truth: docs/data/skilltree.json (the site's Build Explorer renders from it).

Re-confirmed in this pass (2026-08-24):

1. Orientation -- how the tree actually works#

The Last Caretaker's "skill tree" is not a build tree. It is a recipe catalogue with a level gate and a price tag. Nothing you buy makes the Caretaker stronger directly -- it makes an item craftable. The stat layer is a completely separate currency (Enhancer points, ten stats x 100 levels), and confusing the two is the single most common error in community threads about this game.

1.1 The point economy -- and a correction#

A widely repeated shorthand says "one skill point per level." The arithmetic refutes it. The documented cost of the tree is 200 skill points across the 118 unlocks whose price is published; 22 unlocks have no published price. If those 22 average ~2 SP each, the true price of the complete tree is roughly 245 SP. Thirty-five levels at one point each would buy 35 -- about 14% of the tree. So the grant must be several points per level.

What the evidence actually supports:

Practical reading (v1.0 estimate): average skill-point income works out near 7-8 per level and rises with level. Treat that as a planning heuristic, not a published number -- the per-level grant is genuinely unpublished and the wiki has no Skill Points page (HTTP 404, checked 2026-08-24).

✅ SUPERSEDED by the v2.0.0 datamine -- and the estimate above was very close. DT_VoyageCharacterLevels publishes the grant outright: 3 SP at level 2 rising to 9 at level 9, then a flat 10 per level to the cap = 305 SP total, against a 267 SP tree. The "several points per level, rising" argument in this section was right; the 7-8 average was low only because it did not know the flat-10 band runs for 26 levels. The whole of §1.1's reasoning is retained above as the audit trail for how the corpus got there without the data. Enhancer points run on a parallel band table (10/15/20/25 = 700 to the cap) -- see 21-player-builds-leveling.md §3.2.

The two conclusions that matter for how you play:

1. You will finish the tree. No pick is permanently exclusive. There is no respec, but there is also nothing to respec out of -- a "wrong" pick costs you tempo, not access.
2. Therefore ordering is everything and nothing else is. The whole value of a progression path is when you get a capability, not whether. That reframes the entire exercise below: these are tempo plans, not builds.

1.2 The cost curve#

Two independent sources agree on the price of ~30 overlapping unlocks: the current wiki's per-item infoboxes (a cost = parameter, maintained per item) and TheGamer's Complete Guide To Skills (2025-11-08, a launch-era full tier listing). They match on all but two entries.

⚠️ Corpus correction. 05-caretaker-systems-survival.md carries "1 point at Tier 1 rising to 6 points at Tier 21" from the discredited 21-tier SEO guide. No unlock anywhere costs 5 or 6. The observed maximum is 4 (Sabatier Reactor T16, Eclipse T18, Distillation Tower T20, Entanglement Lens + Quantum Duplicator T21), corroborated independently by the wiki infoboxes and by TheGamer ("ranging from 1-4 points per skill"). Treat the 1->6 curve as retired.

1.3 Where the data in §2 comes from#

The wiki's Cargo extension does not carry a skills table -- action=cargotables returns only food_recipes, material_recipes, memories, parts_recipes, professions, and parts_recipes has no tier or cost field. The working route was the plain MediaWiki API: action=query&prop=revisions&rvprop=content against batches of item pages, reading the tier = and cost = parameters straight out of each {{Infobox ...}} call. That yielded machine-read tier/cost pairs for ~120 unlocks in seven calls, which were then cross-checked against the corpus's ~50 pre-verified placements (every single one matched) and against TheGamer's independent cost table.

Confidence tags used in §2:

Provenance labels added in v2.1.0 (used alongside the confidence tags above, and carried in skilltree.json):

Evidence hierarchy, strongest first (the rule this document adjudicates by): player purchase / refusal > in-game tooltip text > screenshot digit read > datamine > wiki infobox > launch-era guide. A datamined value outranks the wiki on structure (tier, cost, requirement class); a player purchase outranks everything on behaviour (whether the button actually works today).

2. THE TREE#

Read the SP column as the price of that single unlock. Tier totals at the end of each block are the documented subtotal only -- unlocks marked ? are excluded from the arithmetic.

⚠️ How to read section 2 after v2.0.0/v2.1.0. Tables 2.1-2.6 are the wiki-era reconstruction, kept intact for audit because it is what every web source says. Rows corrected or priced by the datamine and by play carry an inline [datamined] / first-hand marker and, where a row moved tier, a SUPERSEDED note pointing at its new home. The subtotal and running-total lines are NOT re-derived -- they remain the wiki-era arithmetic (200 SP documented / 203 with the Lazarus Pod). The current economy is the mined one: 267 SP across 154 priced tier unlocks against 305 SP granted by leveling (v2.0.0 note, section 2.8 and docs/data/skilltree.json, which is the machine truth the site's Build Explorer renders).

2.1 Tiers 1-5 (levels 1-5) -- the opening#

Documented subtotals: T1 = 0 · T2 = 5 · T3 = 7 · T4 = 7 · T5 = 8. Band total 27 SP (6 unlocks unpriced).

2.2 Tiers 6-10 (levels 6-10) -- the infrastructure band#

Documented subtotals: T6 = 9 · T7 = 8 · T8 = 10 · T9 = 8 · T10 = 13. Band total 48 SP (5 unpriced). Running total 75 SP.

2.3 Tiers 11-15 (levels 11-15) -- the capability band#

Documented subtotals: T11 = 11 · T12 = 8 (+3 Lazarus Pod) · T13 = 15 · T14 = 12 · T15 = 10. Band total 56 SP (+3 gated; 5 unpriced). Running total 131 SP (134 with the Lazarus Pod).

2.4 Tiers 16-20 (levels 16-20) -- the industrial band#

Documented subtotals: T16 = 16 · T17 = 14 · T18 = 12 · T19 = 5 · T20 = 9. Band total 56 SP (6 unpriced). Running total 187 SP.

2.5 Tiers 21-25 (levels 21-25) -- the tail#

Documented subtotals: T21 = 10 · T22 = 3 · T23-25 = 0 (nothing priced). Band total 13 SP (4 unpriced). Grand total: 200 SP documented across 117 nav unlocks, +3 for the gated Lazarus Pod = 203 SP.

2.6 Budget table (documented cost only)#

2.7 Levels 26-35 -- what the last ten levels give you#

Finding: there are no skill tiers above 25. Levels 26-35 grant points, not content.

The evidence chain:

inferred conclusion: levels 26-35 are budget levels. They pay out skill points (finishing off tiers 1-25) and enhancer points (toward the 1,000 needed to max all ten enhancers -- far more than a playthrough grants). Practically, the last ten levels are where you buy the enhancer stats you have been rationing: Battery, Power Management, Haul Mastery and especially Pressure, which is the only enhancer that gates content.

⚠️ Caveat. U05 shipped a Stainless Steel skill unlock, the Recovery Grinder, the Gyro Deck and the Eden Frame line, none of which appear in the nav template. Whether those live on a separate Project Eden progression surface or the nav simply lags is unresolved -- so "the tree tops out at 25" is safe for the nav-listed tree and UNVERIFIED for the game as a whole.

2.8 Datamine + first-hand corrections to the section 2 tables (v2.0.0 - v2.1.0)#

The audit trail for "what did the mine and the play sessions actually change". Every row is already applied in place above and in docs/data/skilltree.json.