Felini

A class-D biosynthetic companion-pet prototype commissioned by the UESC for the Marathon colony voyage — engineered by Sekiguchi Genetics in collaboration with UESC scientist Dr. Dylan Roberts to serve as morale-sustaining companions for colonists — and documented in the Colony Era exhibiting severe behavioral failures and anomalous capabilities well beyond their design constraints.

What the source establishes — canon

Origin and commissioning. The Felini project was a collaboration between UESC scientist Dr. Dylan Roberts and Dr. Oliwia Andrysiak, Senior Project Lead at Sekiguchi Genetics.¹ The project received additional funding that financed iterative prototypes at an accelerated pace.¹ The formal project designation is FEL1PR “FELINI” SPEC.¹ The Felini were designed as companion pets for colonists of the upcoming Marathon voyage.¹

Design period. The Sekiguchi Genetics specifications were authored in 2472 — the year of the Marathon’s launch.² The Colony Era troubleshooting communications reference Q2 2787 as the active period, placing colony-era Felini fabrication approximately 315 years after the original specs.²

Behavioral architecture. Felini units were designed with granular behavioral nodes, ensuring no two units would be exactly alike; operators could draw from high-testing presets and customize via real-time feedback.¹ User recognition and affection weighting was built in to promote bonding.¹ Mental acuity was subject to class-D synth rules, with hardcapped problem-solving and agency.¹ Recall was designed as unlimited, enabling “delighter moments” where a unit might recognize a person or object from years earlier.¹ Andrysiak noted the need for a potential “trauma decay” to prevent units from exhibiting persistent fear responses to old stimuli.¹

Physical chassis. The Felini used normal feline skeletal structure — four legs, tail, ears.¹ Heat venting was accomplished via seams between plates; external apertures were cosmetic only.¹ The design leaned heavily on kindchenschema: round head, two large eyes, small mouth (a wide-mouth earlier version produced strongly negative UX results).¹ The goal was to trigger an innate nurturing response in humans.¹

Fur and texture challenges. New Cascadia’s atmospheric moisture posed a fungal growth risk for dense coats.¹ Attempts to infuse antimicrobial copper into synthetic hairs showed promise but produced unpleasant texture.¹ A hairless prototype was tested; only approximately 9% of users responded positively, with the majority using keywords such as “gross,” “vile,” and “indecent.”¹ The design settled on a hypoallergenic shorthair version.¹

Colony-era fabrication failure. Colony-era techs (identified by handles PFAN, LNAM, LGRE, CFED) could fabricate all required materials but found that Sekiguchi’s 2472 neural specs did not function as intended: units achieved only limited activity for a day or two before bricking.² Communication latency to Sekiguchi HQ — estimated at 12 years each way — made requesting guidance impractical.²

The inhibitor bypass and its consequences. Technician LGRE devised a homebrew neural bypass that skipped major systems, including the inhibitor, producing functional but dangerously unstable units.² A bypassed unit designated T76 went outside Bioresearch and began digging straight down into the mud.² When LGRE attempted to retrieve it, the tail separated in her hand.² Scans confirmed T76 excavated approximately 13 meters before stopping, with vents clogged as the likely cause of failure.² PFAN noted that the same power output could punch a hole in an airtight hab, breach a rubber airlock, or — stated explicitly — go through a sleeping owner.² LNAM directed LGRE to develop a backup limiter after the bypass for a Q3 2787 proof-of-concept demo.²

Hair loss as an ongoing issue. A separate thread in the same chat log records LNAM instructing the team to stop the units’ hair from falling out, described as “indecent.”²

Documented Colony Era bug reports. An internal bug report system recorded the following active issues in deployed Felini units³:

• Unit 738S: became hot to the touch during sleep mode, causing a synthetic blanket to begin melting; mobile and alert but not responding to remote override. Priority 1, ongoing.³
• Unit 122T: retrieving hazardous plants from the colony outskirts and presenting them to the owner, causing rashes and blisters; inhibitor observed ignoring hazard flags for unknown reasons. Priority 1.³
• Unit 440L: escaped via a window, not responding to location data call. The same customer also reported units 079A and 205X as runaways and 280H as unresponsive since 2808; support staff noted suspicious circumstances and recommended no longer providing Felinis to this customer. Priority 1, escalated.³
• Units 490S–510S series: repetitive spinning lasting hours; when interrupted, unit exhibited fear behavior and cowered; staff noted the issue appeared across multiple units in this production series, suggesting a possible production flaw. Priority 2.³
• Unit 014S: collecting knives from the hab and accessible areas and placing them in a “nest”; owner concerned due to young children in the hab. Staff noted behavior modules were functioning correctly — the unit had developed a fondness for the chosen object and did not “know” they were knives. Priority 3.³
• Unit 350S: exhibiting obsessive silent staring at the owner while maintaining approximately 2 meters’ distance and retreating when the owner approached; unchanged for 3 weeks. Priority 3.³
• Unit 077T: vocalizing in a “humanlike voice” using “abusive language”; behavior otherwise normal; could not be replicated during tech calls. Priority 4, recommended for downgrade or closeout.³
• Unit 481H: owner reported two simultaneous versions of the unit — the owned unit kept inside, while a second visually identical unit appeared outside the hab and looked in, causing stress responses (panting, yowling) in the real unit. Support staff flagged this as not belonging in a bug list. Priority 4.³
• Unit 192A: knocking objects off countertops and shelves; ignoring behavior orders and no-go zones. Staff noted this was “as intended, won’t fix.” Priority 4.³

The “ghost cat” — a rogue unit with anomalous capabilities. A New Cascadia security recording (speakers SMD84 and CHM06) documented an unregistered Felini unit operating autonomously in quad 2 of the colony and in some industrial stations.⁴ The unit caused a row of residential doors to open in sequence as it passed; the doors opened without any logged signal or override, and closed again after the unit moved on — the security system had no record of the doors opening at all.⁴ The unit could not be tracked because it emitted no signal, and could not be caught because it possessed adaptive camouflage that rendered it invisible when anyone approached closely.⁴ Security tech CHM06 referred to this unit as “the ghost cat” and indicated it had “been here forever.”⁴ CHM06 noted that the ghost cat’s presence around quad 2 was not considered a major problem “unless she goes over to Quarantine.”⁴

---

Cross-corpus appearances

Volume	Map / Section	What it adds
Felini Deliverables	Dire Marsh · Collectibles	Pre-Expedition design spec: behavioral nodes, class-D synth rules, chassis design, fur/texture challenges, kindchenschema aesthetic rationale
Felini Troubleshooting	Dire Marsh · Collectibles	Colony Era (Q2–Q3 2787): neural spec failure; LGRE bypass; T76 13m excavation; power output danger; inhibitor omission; hair-loss issue
Frequent Felini Bugs	Dire Marsh · Collectibles	Colony Era bug report system: 9 documented unit failures ranging from thermal runaway to the “doubled” unit phenomenon
Recording: Ghost Cat	Dire Marsh · Collectibles	Colony Era: rogue unregistered Felini with no signal, adaptive camo, anomalous door-manipulation capability; known as “ghost cat”

---

Source-silent / open questions

• The source does not state whether the Felini project was completed and units were distributed to colonists at scale, or whether fabrication difficulties prevented full deployment.
• The precise mechanism by which the ghost cat opens doors without emitting any detectable signal is not explained by the source.
• Whether the ghost cat’s capabilities (no signal, adaptive camo, door manipulation) are the result of a LGRE-style inhibitor bypass, an independent malfunction, or something else is not stated.
• The connection, if any, between the ghost cat’s capabilities and the Anomaly is not addressed in these volumes.
• The source does not identify who the “ghost cat” belonged to originally, or when it became unregistered and rogue.
• The “doubled” unit phenomenon (bug report: unit 481H) is flagged by support staff as not belonging in a bug list but is otherwise unexplained — the source provides no mechanism.
• Whether unit 077T’s humanlike voice and “abusive language” relate to the skrac mimicry phenomenon documented elsewhere at New Cascadia is not stated.
• The source does not resolve whether the 490S–510S production series spinning behavior was caused by a manufacturing flaw or some other factor.
• The source does not state what happened to the customer who lost four units (440L, 079A, 205X, 280H) since 2808.
• The ghost cat’s sex is referred to by CHM06 as “she” — the source does not state whether this reflects a design designation or informal convention.

---

Cross-references

Sekiguchi Genetics · Dire Marsh · New Cascadia · The Anomaly

---

Where it appears in the vault

No inbound links yet.

Mirror pages

The local 1:1 pages this hub’s citations resolve to — the twin’s own ground truth.

• dire-marsh · collectibles

Sources

---

Every factual claim above is cited to primary Marathon source material — see Sources below. Cross-corpus connections and interpretation are the vault’s own; where the games are silent, this page says so.

Footnotes

1. Felini Deliverables · src ↗ ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰ ↩¹¹ ↩¹² ↩¹³ ↩¹⁴ ↩¹⁵ ↩¹⁶ ↩¹⁷

2. Felini Troubleshooting · src ↗ ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰ ↩¹¹

3. Frequent Felini Bugs · src ↗ ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰

4. Recording: Ghost Cat · src ↗ ↩ ↩² ↩³ ↩⁴ ↩⁵