Project P.I.T.T. Speedrun Guide: Any%, Secret Ending, and Fast Factory Routes

What a Project P.I.T.T. speedrun actually is

A project p.i.t.t. speedrun is not a traditional level race. Project P.I.T.T. (also written Project PITT) is a first-person incremental physics sandbox from Froke / Fr0ke and Pretty Soon. You play an intern in a brutalist plant built around The Maw. The loop is always the same: produce items, dump them into the pit, keep a combo alive, cash out, and buy better machines.

That loop is exactly why the game speedruns well. Time is decided by how fast you open the shop, how quickly you turn a messy duck pile into a reliable lane, and whether you chase the first credits sequence or the longer upstairs secret chain. Movement tech, physics flight, and factory design matter as much as knowing which ending you want.

The full game launched on Steam on 19 August 2026 for Windows, Mac, and Linux. If you only want to practice movement and the first production cycle, the official demo on Steam is the legitimate way to try Project PITT before buying.

Official speedrun categories

Community boards on Speedrun.com already split Project P.I.T.T. into short joke categories and full-game Any% runs. Treat the names below as community reports from the public leaderboard, not as developer-published rules.

Main Ending and Secret Ending are the two Any% branches that matter for a serious project p.i.t.t. speedrun. SingleDuck and Gamble% are useful warm-ups. They teach throw timing and crate routing without asking you to sit through a multi-hour production grind.

Current community records

These times are a late-August 2026 snapshot from public community boards. They can move quickly on a new game, and they should be read as player reports, not official developer rankings.

The secret-ending mark being faster than the older main-ending mark is not a contradiction by itself. Community reports list different patches, and the board is still tiny: only a handful of players and a little over ten submitted runs at the time of that snapshot. If you submit, expect vanilla versus modded to be separated, and expect verifiers to care about game version.

How the factory loop decides your time

Money per duck is a trap. In a Project PITT speedrun, the clock cares about throughput. A combo grows when many products hit The Maw in a short window. The chain dies if you leave a long gap. Later items and piñata candy are worth more than starter rubber ducks, but a clogged Automater still loses more time than a cheap duck lane that never stops.

That means a good Any% route is a sequence of unlocks, not a shopping list:

1. First duck into The Maw to open progression.
2. Throw upgrade so you can feed without walking every item to the rim.
3. Cash Register as soon as the terminal offers it.
4. Automater on immediately, aimed into a lane, never at your own feet.
5. Sticky tools so panels stay put.
6. A short funnel: producer, stuck panels, fan, Maw.
7. Only then do you scale into magnets, piñatas, and late assembly parts.

Standing in the Automater output is a classic reset. Your body becomes a dam, the combo dies around 70–80, and you spend a minute broom-cleaning a problem you created.

Fast early-game routing

The casual 100% path and a speedrun path start the same and then diverge.

Shared opening

• Craft one duck and feed it. That is QWACK! in a normal playthrough and the entire SingleDuck category in a race.
• Use the press until the first shop wave appears.
• Buy throw, then either the broom or the first fan. The broom is safer for spills; the fan is faster if you already know a funnel shape.
• Buy the Cash Register the moment it appears.
• Turn the Automater on and start a pile, then dump it in one burst for the first large combo.

Where speedruns skip the tourist route

A completionist pauses feeding to leave 500 and then 1,000 loose products on the floor for mess-related achievements. A runner does the opposite. Loose product is lag, lost combo, and visual noise. You only leave a pile on purpose when you are about to magnet-dump it.

Gambling crates are the one early side objective that still matters. The related achievement is tied to finding them, and unidentified boxes should not be yeeted into The Maw until you have checked them. In Gamble% that hunt is the run. In Any%, crates are optional money and special ducks, not your long-term economy. Community reports treat them as a bonus, not a primary income engine.

Physics movement that saves minutes

Project P.I.T.T. wants you to abuse its sandbox.

Jumping while holding an object is an intentional flight trick. Runners use it to reach catwalks, fuse boxes, and the upstairs scale without building a staircase. Mouse wheel changes hold distance, middle mouse rotates freely, R snaps rotation, and G resets it. Those bindings are not flavor. They are how you keep a held crate stable while you “fly” a gap.

Height has a tax. Climb too far and a five-second countdown can end in I SEE YOU, a surveillance reset. The counterplay is to smash cameras with thrown products or to use the lights at the ends of support pillars to disable the height lock. Secret-ending runners have to learn this immediately, because the upstairs route is mandatory for them and optional time for Main Ending.

If you clip into geometry, use Unstuck in the pause menu. Resetting a save in a timed run is almost always worse than a two-second unstuck.

Factory layouts that scale for Any%

You will rebuild. Later facility phases drop scripted debris onto whatever pretty line you just finished. Do not marry a layout. Rebuild with the new tools and, if the fans are strong enough, shove some of the rubble into The Maw.

For a main-ending project p.i.t.t. speedrun, the standard lane is the backbone. Magnet dumps are the burst tool when you need a shop unlock now. The golf club is a bridge, not a strategy. Once fans and upgraded producers beat a swing, put the club down.

Aim producers into the lane, not the floor. Keep the route short. Clogs kill more runs than “wrong” item value. Community reports also say performance itself becomes a speed mechanic: if you ever farm a thousand loose items, clean them before you try a high-throughput segment, and drop retro effects or resolution if the frame time starts eating inputs.

Main ending versus secret ending

The two Any% boards are different games wearing the same opening hour.

Main Ending is the first credits-style sequence. You keep feeding The Maw until the first statistics dump appears. You still buy aggressively, because better machines are how you stop the clock, but you can ignore most of the upstairs mystery if it does not pay for itself.

Secret Ending is the full upstairs chain:

1. Reach the catwalks with object flight or improvised climbing.
2. Find three fuses and seat all of them. Climbing alone does not count.
3. Load the 0/200 kg platform with mass, not item count. Gambling crates, bumpers, remotes (about 1 kg each), or a panel box packed with heavy junk all work.
4. Use the newly opened grate, door, or chute.
5. Feed the numbered percentage pipe with later-phase or assembled products. Raw starter ducks leaving the pipe at 0% is expected, not a bug.
6. Keep buying, assembling, and feeding through the remaining facility phases until the final ending unlocks.

That is why secret routing can look “slower on paper” and still post a faster time on a newer patch: the upstairs work is extra, but later products and a solved pipe can print money that a main-ending factory has not unlocked yet. Until more vanilla same-patch races exist, treat cross-board comparisons as community observations.

Hidden surveillance beats (BAD?, INVISIBLE?) are achievement bait. They are not required for either Any% board unless a specific ruleset says otherwise. Do not stare at the observation window unless you are going for 100%.

Short categories worth practicing

SingleDuck is a 30-second exam. Spawn, make the first duck, throw it cleanly into The Maw. Learn throw upgrade timing only if the category allows pre-unlock; otherwise it is a movement and release-angle test.

Gamble% is a scavenger hunt. Community write-ups say crates hide on the main floor, catwalks, delivery pipes, dark rooms after you restore lights, and upstairs. Do not dump unmarked boxes. The whole point of the category is that you found them.

Demo is the honest practice board. Use it to rehearse flight, camera breaks, and the first Automater lane without burning a full-game attempt.

A practical Any% checklist

Use this as a runner’s abbreviated route, not a 100% tour.

1. First duck, shop open, throw upgrade, Cash Register.
2. Automater plus a one-fan funnel. Get a x100 dump, then a x200 dump once the pile is real.
3. Sticky tools as soon as panels exist. Rebuild after every phase change instead of fighting debris.
4. Buy every new terminal unlock the moment it appears. Waiting for a “better time” is how you stay on ducks too long.
5. Unlock Piñata and Anomaly when they show up; candy and later oddities out-earn starter product.
6. If you are on Secret Ending, do fuses and the 200 kg scale as soon as flight is comfortable, then come back with assembled products for the pipe.
7. Push a short, clog-free lane toward high items-per-minute. Sustained rate beats a single heroic burst.
8. Stop the clock on the ending your board asked for. Do not accidentally tour the other ending unless the ruleset is “any ending.”

A completionist still needs 15,000 lifetime feeds, a x300 combo, one trillion dollars, and every terminal purchase. Those are real goals, and they teach factory skill, but they are not the Any% finish condition. Know which clock you are racing.

Common time losses

• Combo death at 70–80. Look for leaks, a fan pointed at a wall, or your hitbox in the shaft.
• Phase change deleted the factory. Scripted. Rebuild. Stronger fans can eat some of the wreckage.
• I SEE YOU on every climb. Break cameras or shut the pillar lights off before the next flight.
• Pipe stays at 0% after 200 kg. You fed the wrong generation of product.
• Nine of ten crates found. Recheck pipes, dark rooms, and upper walks. Community reports keep losing the last one there.
• Useful puzzle piece fell in. Unstuck if you clipped; buy a replacement if the terminal still sells it. Never feed fuses on purpose.
• Mod confusion. If you care about the vanilla board, do not install quality-of-life packs that the ruleset would mark as modded.

The public board is still young: a few moderators, a small Discord-linked scene, and room for routing to change after the next patch. Read the current rules on the leaderboard before you record. Versions already split times.

FAQ

Is a Project P.I.T.T. speedrun the same as 100%? No. Any% stops at the main or secret ending. Full completion still asks for huge lifetime feed counts, a x300 combo, a trillion dollars, every shop purchase, and the crate hunt. Practice the factory on 100% saves, then strip the route down for a race.

Do I need mods to be competitive? Not on the vanilla board. Community reports already show a sub-3h20 main-ending run listed as vanilla. Mods may be legal on a separate column; check the category rules before you submit.

Where should I practice if I do not own the full game? Use the official Steam demo (or the itch demo if you prefer that storefront). There is no legitimate full-game substitute. SingleDuck, flight, camera breaks, and the first Automater lane all transfer from the demo into a real project p.i.t.t. speedrun.