Hold The Horde — Whole Game Outline

You are an astronaut stranded at an unofficial station whose only message says stay away. Your ship can keep you alive, but cannot leave. Beyond its airlock are limited oxygen, crowds of former patients and machines built by two vanished cultures. When you die, you wake aboard with everything physical you were still carrying. The unjoined station rearranges itself; doors you opened and room connections you deliberately made remain.

The campaign grows from immediate survival into a larger question: repair the ship, learn why this hospital restores the dead, decide what—if anything—to change at its core, and perhaps turn every shifting room into one place you built yourself. “Apparent,” “deeper” and “optional” are discovery layers, not a fixed hierarchy of endings.

The rhythm of an expedition

At home, choose a connected equipment rig and decide how much pack space to give oxygen instead of tools or findings. In the station, ordinary weapons fire mostly automatically; the player’s attention stays on movement, routes, crowd shape and manually timed special abilities. Fighting regularly grows temporary power, and major horde moments can transform the rig into something ridiculous before the fight is over.

Discovery happens inside that pressure. Photographing a mark, fitting a new part and operating a machine all happen while enemies and oxygen remain live. Opening the journal pauses the entire world, including enemies and oxygen, so the player can compare records and write their own interpretation without a reflex test. The journal preserves evidence, not truth or current routes.

Every return home ends the temporary power climb and prepares eligible combat areas for fresh hordes. A living return keeps the station’s arrangement; death also shifts its unjoined parts. In either case, carried equipment, components and loot—including rig-mounted parts—come aboard. Breathed oxygen and consumed materials stay spent; objects deliberately left in station or ship fixtures, dropped or handed over stay put. Permanent work, seals and selected joins persist. The next trip should offer two reasons to leave: try a changed machine and follow the last discovery.

Opening — one part, two machines

The first goal is brutally practical: stop the broken ship from damaging itself. From the dock, the astronaut reaches Cargo Turn and finds a loose Fork Relay, a component that can split one trigger between two compatible receivers. In the adjacent Collar Control space, the isolation bridge needed by the ship is still fixed in place. Removing it is noisy work, and a patient-bay shutter can release a crowd into the route home.

The Fork can serve the rig or the station board. Leave it in the board and the rack and shutter can remain powered together, buying a safer extraction but leaving the part behind. Keep it and the player carries the first flexible build component home, at the cost of pack space and a more exposed fight. A simpler selector still allows the bridge to be removed, so progression does not require the “correct” loadout.

The intended pleasure is the first conceptual surprise: a “key” and a combat part obey the same connected-machine grammar. The player returns with a real ship repair and an idea they can experiment with. This phase could fall flat if it reads as a wiring tutorial with enemies waiting politely, rather than a dangerous decision about what to carry and what to leave operating. Option A is selected as the opening direction; exact timing, quantities and final room presentation remain design work.

Early expeditions — a person, a language and a place worth remembering

The astronaut next wants a dependable foothold and proof that somebody else survived. Oxygen makes every detour compete with the return. A short current route to a quarantine refuge can support one cooperative rescue: the player walks and prepares the route, supplies travel air and holds a single contested junction while the survivor moves under their own power. It must not become a long escort. Which person is first, and whether a death during the crossing ends at shelter or aboard the ship, remain open choices.

The rescued person changes home through a practical contribution, a personal lead and an imperfect account of the hospital. They do not become a translation device. A survivor responds only to evidence the player actually shares, may remember a crew use or nickname, and can disagree with another person without withholding the main game. Preparing oxygen, changing equipment links and deciding what testimony to pursue now happen in a growing home rather than an empty loadout screen.

The player also begins recording two genuinely different alien languages. Builder marks recur on facilities and possible access panels; a later visitor hand marks transfer relationships and responsive work lines. The current SMALL map contains four language discovery chains across those two languages. Four separate story or personal investigations reuse some of the same records, making eight chains in the campaign crosswalk—not eight languages or eight disjoint routes. Each useful reading predicts an action that can be tested physically; no required objective depends on receiving a translation.

Death makes an old corridor record historical, but the pictured room, mark and completed work remain worth recognising when that room appears among different neighbours. This phase could fall flat if rescue becomes a preparation checklist, survivors feel like upgrade benches, or the scripts reduce to matching symbols to highlighted buttons.

Middle expeditions — fight your way to a revelation

The middle campaign opens several priorities at once: repair more ship functions, recover new equipment, bring a small cast home, investigate patient transfers and preserve useful room connections. The player can follow build opportunities before language, use language to shorten a dangerous operation, or postpone a rescue until its current route is genuinely short. No single survivor, translation or death roll is a universal gate.

One current expedition shows how the promises meet. In the Cable Junction, points of light above a dark void move when a parked rail lamp twitches. A folded bridge aims into the darkness. The player releases the lamp carriage and discovers that its motion wakes visible service runs behind and beside them. Bodies descending the Pump Hall gantry and pushing through the junction’s duct can stall the rail. Automatic fire keeps working while the player runs ahead, circles for an angle, clears the track or pulls the brake and retreats.

With the Storm Choir rig, gun impacts feed arcs and charge a store. Breaking ordinary pressure offers Echoing Arc; answering the Veilbearer while keeping the scan moving can ready Thunderhead. At the crest, a paper target physically overlaps 70–120 bodies from 100–200 available across three announced sources—not 200 at once, adaptive refill or tested performance—so the player risks the centre and manually releases the storm to clear the lamp’s path. Neither upgrade is a key: ordinary fire, cutter and retreat remain valid.

When the noise finally drops, the apparent stars resolve into receiver fittings seen from three positions. One aligns with a collar at the Regulator Gallery. The player releases and walks a bridge across the void. With little oxygen left, they can cross immediately or return with the record and prepare. A living return preserves the aligned bridge as a shortcut; a death-shifted arrangement makes the old samples historical and requires a visible rewind before a new scan. Combat has bought quiet, the observation has become a route, and the next expedition starts from a changed question.

A different middle trip gives discovery a human consequence. Builder marks and visitor transfer pairs can predict which pressure control to try while a horde contests the valve. Under the found-part proposal, take the sampling loop’s Impact Dynamo for a rig and lose its local impact response, or leave that station function intact; a regulator component serves the ship. A ledger proves only that one transfer happened later than a survivor believed. Back home, Yeva may revise one accusation while Dorn keeps another interpretation. The game changes a machine, a route and a relationship without declaring who was right about the station.

The current Storm Choir, Ricochet Foundry and Gravity Furnace rules live in the Equipment Catalogue. Build Comparison offers different Ricochet and Gravity fantasies, but those proposals have not replaced the shared equipment rules. This phase could fall flat if renewed hordes make familiar travel feel like compulsory warm-up, the moving lamp becomes escort duty, or combat warnings and machine cues crowd out the revelation.

An apparent way out — leave, or keep asking

Four recovered ship assemblies can produce a visible readiness report. This gives the astronaut a credible practical achievement without requiring a language solution, a survivor or the core. The important player question is emotional as well as mechanical: is repairing the original escape enough, or has the station become something they cannot leave unexplained?

One proposed apparent ending makes undocking an explicit, irreversible choice that ends revival and station exploration. The player can preview that price, depart for a coherent close, or postpone indefinitely. The alternative keeps readiness as a repair milestone without selecting departure or credits yet. Neither is owner canon. This layer could fall flat if readiness is only a checklist and leaving has no human or mysterious cost worth considering.

Deeper investigation — act without possessing the whole truth

Operational knowledge, physical alternatives and evidence from people can eventually bring the astronaut through the core’s pressure iris. Two alien languages may make the correct control predictable, but gauges and a slower two-clamp route keep the action physical and available without translation. Behind it, the hospital story remains deliberately revisable.

The leading proposal places living tissue behind the isolation boundary; the strongest alternative leaves an empty template engine. In one branch, re-seating isolation may disconnect a harmful mixed profile while leaving ordinary revival intact. In the other, the question concerns which repair profile a machine should use. These are team proposals, not settled history or endings. The owner has not decided whether corruption ever changes the astronaut, whether voluntary treatment is used, or what the core ultimately contains.

The player may inspect, intervene, leave the mechanism alone or return with better evidence. People aboard react only from what they know and what the player shares. A mechanical response does not prove a speaker, living tissue does not prove personhood, and an empty cell does not explain every event. This layer could fall flat if one dramatic image answers the mystery for the player, or a reversible control feels like a settings switch with no consequence they care about.

Deepest optional play — make the station one place

Early persistent joins are useful: keep a strange machine near a known route, preserve two rooms whose evidence belongs together, or create a safer return. After death, joined rooms remain together while their unbound outer neighbours may change. The player gradually realises that these local acts form one of the deepest puzzles: connect every playable room in the chosen scope into one continuous station rooted at the ship.

This is meant to turn disorientation into recognition, competence and finally ownership. The player must recognise rooms, test compatible opportunities and preserve a route they chose; an old journal record never grants current access. Deliberate death may expose a new arrangement, but the design must offer several useful leads and must not require grinding one exact adjacency.

What the final join does is still open. One proposal makes the completed station a dependable place whose internal geography no longer changes. Another adds the option to move and eventually depart with the whole station and docked ship, at substantial new story and production cost. Both preserve continued hordes and a fresh temporary climb until any separately chosen departure. This layer could fall flat if joining becomes a long checklist, the complete graph cannot be assembled fairly, or stability removes geographical surprise without leaving strong investigations to finish.

Current scope reference

The current detailed planning reference is SMALL: four regions, 24 reusable room modules, 40 placed rooms, 12 expedition objectives, five people with 15 primary scenes, 12 equipment definitions, ten encounter placements including two setpieces, five mechanism families and 14 proposed major placed mechanisms. It also carries the four language chains and four story/personal chains described above.

This is a reference, not an owner-selected game size. The reviewed Campaign Scope Choices keeps a credible 36-room/22-module concentration open; it removes spatial variety while retaining the headline campaign. Larger TARGET and STRETCH bands remain ambitions, not commitments. None of these counts demonstrates 30, 90 or 100–200 hours of worthwhile play.

The historical 350–590 role-hour allowance covers further detailed-campaign documentation for an earlier 13-mechanism basis. It excludes the existing first pass, implementation, assets and testing. The fourteenth mechanism—the moving scan and bridge—and its revised placement geometry remain uncosted, so the old range is not a current total, remaining-work estimate or ceiling. A room cut does not justify a proportional effort reduction.

Choices that remain open

The hospital truth: living captive, empty template engine or another revision; every causal detail remains revisable.
Whether ship readiness leads to an irreversible apparent departure, and what any deeper or “real” ending means.
Whether and how corruption affects the astronaut; no death counter, forced transformation or treatment choice is selected.
The first survivor, rescue failure outcome, post-rescue mortality and exact colony-management scope.
The final joining mechanism, a valid full layout and what completion does to later shifts.
The content band, exact combat-growth cadence, acquisition model, oxygen quantities and final equipment rules.

Open Decisions keeps the current decision list. Detailed ownership sits in Hordes and Builds, Fighting the Horde, Two Language Discovery Map, Story And Ending Map, Discovery in a Changing Station, Completing The Station, Production Scope and Campaign Scope Choices.

This outline describes intended experiences and design hypotheses. It does not establish fun, readability, balance, feasibility or campaign duration.