# Litter Products: research notes Date: 2026-10-25. Scope: the litter itself — the granular material poured into a cat's litter box or tray to absorb urine and bury waste (clumping clay, silica-gel crystal, and plant-based/biodegradable fillers such as corn, wheat, walnut, pine, paper, tofu and grass). Overwhelmingly a cat product; the register is written so its facets would also describe litter sold for rabbits or ferrets in an enclosed habitat, but every sourced claim below is about cats specifically, and one judgement (J4) flags a cross-species hazard rather than extending scope to small-animal bedding. Not in scope: the box or tray itself, scoops, litter mats, box liners, and waste-disposal systems (bags, pails, odor-sealing devices) — these have no dedicated sibling leaf in `catalog.json` yet (`waste-management` is the nearest, status "planned," and historically covers disposal gear, not the box); if that leaf is written later, it should take box/scoop/liner/disposal products, and this register should not be read as covering them. Puppy/dog indoor litter trays exist but are a minority use case not covered by any source below. Quote labels: **verbatim** = I read the sentence myself in text I pulled directly off the server (curl, with tags stripped by a small script) and it appears in the saved `.txt` file in this run's scratchpad folder exactly as quoted. **fetched excerpt** = the WebFetch tool returned this as a quote or close paraphrase from the live page, but I did not independently see the raw page text (its output is a model summary of the fetch, not the page itself). **search excerpt** = from a search-engine result snippet. A search-excerpt-only number does not go in `category.json`. WebSearch reported its session budget exhausted (200/200) on my first three queries, before I had used a single search myself this run, so every source below was reached by direct URL fetch (WebFetch or curl), never by search. ## Seed Seed script output, recorded verbatim: ``` === Example entry (from an unrelated category) === category: kitchen-knives what this shows: a numeric spec that only makes sense as one side of a trade-off, and a visual grammar borrowed from the hobbyist community that puts the trade-off on one axis. Trade-off: optimise steel_hardness, give_up toughness and ease of sharpening. === Persona === An audio engineer, who thinks in measurable versus perceived quality and where marketing lives in the gap ``` 1. What is litters' equivalent of `steel_hardness`? Dust level. Bags carry a front-of-pack claim ("99% Dust-Free", "low tracking") that only makes sense next to what you trade for it — tighter, more aggressive clumping and odor-sealing tend to come from clay that also runs dustier, the same shape as HRC trading edge retention against toughness. The difference, and the thing worth keeping from the persona's question below, is that HRC is an independently measured, universal number; nothing I found gives dust level that status. 2. What would the audio engineer notice first that a typical buyer would not? That "dust-free," "7-day odor control" and similar front-of-bag claims are marketing copy with no independent, standardized test behind them, the way a "warm" amp sound is a real perception with no agreed number attached. THD has a meter; cat litter dust does not. Kept: `dust_level` is written as a listing-stated *claim*, not a measured value, with the facet's own backfill note saying so, and no manufactured day-count or percentage appears as a sourced range anywhere in the JSON. Dropped: I considered giving `dust_level` or odor duration a numeric range (days, percent) pulled from brand marketing copy, and dropped it — a single bag's "7 day odor-free" claim or "4-5 lb for 30 days" owner anecdote (see Trimmed detail) is not a measurement I can stand behind with two sources. ## Sources and what each supports **Wikipedia, "Cat litter"** (primer) — `https://en.wikipedia.org/wiki/Cat_litter`, accessed 2026-10-25, raw text pulled with curl and tag-stripped into `wiki_catlitter.txt`. - verbatim: "Edward Lowe accidentally discovered that clay made an excellent cat litter material and began selling it in bags as 'Kitty Litter' in 1947." Supports: mechanism history. Confidence high. - verbatim: "In the US, cat litter is a $2-billion industry consuming five billion pounds (2.3 Mt) of mined clay annually." Supports: clay-mining scale, used in the mining trade-off. Confidence high (single source for the exact figure; used only as context in notes, not as a JSON number since I have no second source for it). - verbatim: "Subsequently in America, clumping bentonite was developed in 1984 by biochemist Thomas Nelson" and "are made from granulated bentonite clay, which clumps together when wet and forms a solid mass separate from the other litter in the box." Supports: `litter_material` mechanism (clumping clay family) and `clumping` facet. Confidence high. - verbatim: "Non-clumping cat litter is often made of zeolite, diatomite and sepiolite." Supports: mechanism note that non-clumping clay is a distinct, now-minority sub-case folded into the `clumping` facet rather than given its own split value. Confidence high. - verbatim: "silica dust, which in California is treated as a known carcinogen under Proposition 65." Supports: judgement J1 (`crystalline silica`, `dust_level`). Confidence medium — single source; see Corrections/limits below, I could not reach OEHHA's own listing page (empty fetch) or a California-government mirror of it to confirm independently. - verbatim: "Because of the clumping effect, the manufacturers usually instruct not to flush clumping litters down the toilet, because it could clog it" and "Other cat owners can be attracted to the biodegradable litters because of their flushability." Supports: `flushable_claim` facet (which material family carries the claim) and its link to judgement J2. Confidence high for the clumping-clogs-pipes instruction; the biodegradable-flushability sentence is presented by Wikipedia as a motivation, not a safety claim, so I treat it only as evidence that the claim exists on the market, not that flushing is safe. - verbatim: "No replenishing is necessary with silica gel (crystal) litter," contrasted with clay/clumping litter needing "20–30 lb (9.1–13.6 kg) [citation needed] or more" topped up over the same period. Supports: `clumping` facet (why crystal is a clean "no" on the split) and the maintenance-style trade-off. The clay-replenishment figure itself is `[citation needed]` in Wikipedia's own text, so it is not used as a sourced number — only the qualitative "crystal needs no replenishing because nothing is scooped out" point is kept. - verbatim: "this sort of litter can be toxic to ferrets, leading to both respiratory and digestive problems." Supports: judgement J4. Confidence medium, single source. - verbatim: "Guar gum was the principal clumping agent used in biodegradable cat litter as of 2013, with a clumping rate of 69.8%, while xanthan gum (73..." [figure cut by my grep window, full sentence continues "...xanthan gum (73.1%) was superior, and plantago gum (57.5%) were identified in one study as potential alternatives."] Supports: the general fact that plant-based litters need an added gum to clump at all, used qualitatively in the trade-off entry; the specific percentages are not repeated in `category.json` (single academic paper, reached only through Wikipedia's citation, not fetched directly) and are kept here only as colour: guar gum ≈ 69.8%, xanthan gum ≈ 73.1%, plantago gum ≈ 57.5% clumping rate, from Vaughn et al. 2013, *Industrial Crops & Products* 43, as cited by Wikipedia. - verbatim: "Some owners praise its absorbency because 4–5 pounds (1.8–2.3 kg) can absorb liquid and odor for up to 30 days for one healthy normal weight cat." This is explicitly framed by the article as an owner anecdote ("some owners praise"), not a measurement, and it is the one Wikipedia sentence I am treating as a search/forum-grade claim rather than a fact — dropped from the JSON entirely, kept here as colour only. **Wikipedia, "Sea otter"** (primer) — `https://en.wikipedia.org/wiki/Sea_otter`, accessed 2026-10-25, raw text pulled with curl into `wiki_seaotter.txt`. - verbatim: "Urban runoff transporting cat feces into the ocean brings Toxoplasma gondii, an obligate parasite of felids, which has killed sea otters." Supports: judgement J2. Confidence medium, single source (see Corrections). - verbatim: "Necropsies of dead sea otters indicate diseases, particularly Toxoplasma gondii and acanthocephalan parasite infections, are major causes of sea otter mortality in California." Supports the same judgement with a second sentence from the same article (not a second source, so still counted as one). **Ohio State University Indoor Pet Initiative, "Litter Boxes"** (academic — a university veterinary-college outreach program) — `https://indoorpet.osu.edu/cats/basic-indoor-cat-needs/litter-boxes`, accessed 2026-10-25, raw text pulled with curl into `osu_litter.txt`. - verbatim: "Research has shown that most cats prefer fine-grained, unscented litters." Supports: `scented` and `texture_size` facets, judgement J3. Confidence high. - verbatim: "Fill boxes with enough litter to permit the cat to scratch and bury waste (usually about 2 inches deep)." Supports: judgement J3 (kept as depth context in notes; depth itself is not a facet here because it is a box-filling practice, not a property of the litter product, so it stays out of `category.json`'s facet list and is only used to corroborate the texture/scent judgement). - verbatim: "The Golden Rule is 'one litter box per cat, plus one'." and "the litter box must be scooped daily and washed weekly with mild dish detergent." Supports: `multi_cat_formula` facet's rationale (more cats, more boxes, more odor load) and general colour; not asserted as litter-product facts themselves. **ASPCA, "Litter Box Problems"** (primer — national animal-welfare organization's behaviour advice) — `https://www.aspca.org/pet-care/cat-care/common-cat-behavior-issues/litter-box-problems`, accessed 2026-10-25, raw text pulled with curl into `aspca_litter.txt`. - verbatim: "Cats generally prefer clumping litter with a medium to fine texture." and "They also usually prefer unscented litter." and "Most cats prefer clumping, unscented litter." Supports: judgement J3, `texture_size` and `scented` facets. Confidence high, second independent source agreeing with OSU. - verbatim: "Cats usually prefer one to two inches of litter" and "Provide one to two inches of litter rather than three to four inches." Corroborates OSU's depth figure (kept as colour, not a facet). - verbatim: "Make sure you have one for each cat in your household, plus one extra." Corroborates OSU's "plus one" rule. **Whisker (Litter-Robot's parent brand), "Lit Premium Cat Litter" product page** (spec_sheet) — `https://www.whisker.com/greatlitter`, accessed 2026-10-25 via WebFetch (fetched excerpt — this is a JavaScript-rendered storefront page; curl returned only the shell, so I could not pull raw text myself and I am labelling every quote from it as **fetched excerpt**, not verbatim). - fetched excerpt: "One ingredient: sodium bentonite clay... locks in odors fast and clumps tight" and "Sun-dried = low dust, low tracking, softer on paws... No clay litter is completely dust-free." Supports: `litter_material`/`clumping`/`dust_level` facets — the "no clay litter is completely dust-free" line is the clearest industry admission I found that clay dust claims are relative, which is why `dust_level`'s values read as marketing claims rather than measured zero/non-zero states. - fetched excerpt: "Naturally unscented, which better contributes to odor control than perfumed cat litter" and "Best litter for Litter-Robot®... developed to work flawlessly with Litter-Robot." Supports: `scented` facet's framing (a brand explicitly markets unscented as functionally better, not just a preference) and the general point that self-cleaning/robotic boxes are designed and marketed around clumping clay — I could not find or reach an explicit page saying crystal or pine litter is *unsuitable* for Litter-Robot (the obvious candidate URL 404'd and the comparison-chart page, also fetched, did not mention litter compatibility at all), so I dropped a planned `self_cleaning_box_compatible` facet rather than assert an incompatibility I only have one-sided brand evidence for. See Trimmed detail. **Dr. Elsey's, FAQ page** (spec_sheet — veterinarian-founded litter brand) — `https://www.drelseys.com/pages/faq`, accessed 2026-10-25 via WebFetch (fetched excerpt, same JS-rendering caveat as Whisker above). - fetched excerpt: "Quartz is naturally mined and present in all our clay litter. Some of the pieces are larger than others but are not harmful" and "No, our litter does not contain crystalline silica dust." Supports: the `dust_level`/crystalline-silica judgement from the brand side — a second, independent commercial source treating "contains no crystalline silica dust" as a claim worth making, which only makes sense if some clay litters do carry it, corroborating Wikipedia's Prop 65 line qualitatively (not a second source for the Prop 65 listing itself, which remains single-sourced). **Arm & Hammer, cat litter page** (spec_sheet) — `https://www.armandhammer.com/cat-litter`, accessed 2026-10-25 via WebFetch (fetched excerpt, same caveat). - fetched excerpt: "Clump & Seal™ Multi-Cat, Complete Odor Sealing Clumping Litter," "powerful odor eliminators and ARM & HAMMER Baking Soda in every scoop," "a 7 day odor-free home guaranteed." Supports: `multi_cat_formula` and `odor_control_mechanism` facets (baking soda as a named, real odor-control technology, and "multi-cat" as a genuine marketed sub-line, not a facet I invented). The "7 day" guarantee is marketing copy from one brand and is not used as a sourced duration anywhere in the JSON. ### Source types I could not reach WebSearch reported its 200/200 budget used on my very first three queries this run, so all research past that point was direct fetch only, per the brief. Within that constraint: - **Regulator**: OEHHA's Prop 65 list page and chemical page both returned empty bodies to WebFetch and a 212-byte stub to curl (JS-rendered, nothing to parse); `p65warnings.ca.gov`'s fact-sheet page is behind Incapsub/Incapsula bot protection (curl got an Incapsula challenge frame). NOAA Fisheries pages on sea otters and toxoplasmosis 404'd under every URL I guessed, and `cdc.gov`'s toxoplasmosis page returned "Access Denied" to curl (Akamai) though WebFetch itself returned a usable fetched-excerpt summary of it, which I did not end up relying on in the JSON. California Department of Fish and Wildlife's sea-otter page loaded but does not mention toxoplasmosis or litter at all. - **Standards body**: none found. There is no trade or standards association specifically for cat litter (unlike tyres' treadwear grading or lamps' NGLIA); the closest analogue, Proposition 65's chemical listing, is a regulator list, not a trade standard, and is single-sourced as noted above. - **Specialist review**: The Spruce Pets, Catster and a guessed PetMD URL all returned 403 Forbidden or 404; Claude Code's WebFetch refused `thesprucepets.com` outright as a blocked domain. No Internet Archive snapshot existed for the Spruce Pets URL I tried (`archive.org`'s availability API returned an empty `archived_snapshots`). - **Enthusiast community**: Reddit (`reddit.com/r/CatAdvice/wiki/litter`) is a domain WebFetch refuses outright; TheCatSite forum returned 403. I did not find a working substitute within budget, so this register has no enthusiast-community source, which I am flagging rather than papering over with a weaker citation. Net effect: the register's safety-adjacent judgements (J1, J2, J4) are each single-sourced to a Wikipedia article, and I was not able to upgrade any of them to two independent sources despite trying four to six plausible regulator/primary URLs per claim. I am stating this plainly rather than either suppressing the facts (they are real and relevant, especially J2) or dressing them up as better-sourced than they are. Per house rule, no single-sourced number went into `category.json`'s numeric fields — all three single-sourced judgements are qualitative claims in the `judgements` array, not facet ranges, and each judgement's own `note` field repeats the single-source caveat so a reviewer sees it without opening these notes. ## Data points used for derived metrics `cost_per_lb` depends only on `price` and the generic `sizing.weight` field already defined in `supplies-standard.jsonc` — both are arithmetic, not researched claims, so no citation is needed for the formula itself. No other derived metric is included: I considered a cost- or litter-per-month figure (clay needs topping up, crystal and plant-based allegedly last longer per pound) but the only concrete consumption-rate numbers I found were the `[citation needed]` clay figure and the "some owners praise" crystal anecdote in the Wikipedia cat-litter article, both flagged above as too weak to use, so I dropped it rather than invent a plausible-sounding default. ## Against the existing add-on schema Add-on read: `supplies/pets/add-ons/litter.jsonc` — two fields today, `litter_material` (enum: clay_bentonite, silica_gel, recycled_paper, corn, wheat, walnut, wood, tofu, grass, na, unspecified) and `clumping` (boolean, commented "invisible; the litter filter") — composed with `pets-standard.jsonc` (animal_type, sizing.suits_weight_range / max_pet_weight, both irrelevant to a consumable filler and correctly not inherited here) and `supplies-standard.jsonc` (`high_signal_markdown`, `sizing.size_stated/pack_count/dimensions/capacity/weight/measurements_text`). | Facet in this doc | In add-on today | Proposed kind under AGENTS.md | Note | |---|---|---|---| | litter_material | yes, `litter_material`, but granular (9 named materials) where this doc groups by mechanism family | vertical anchor | The add-on's granular enum is still the right storage shape (buyers do care which plant it is), but the register's 3-way family split (clumping clay / crystal / plant-based) is the cold-start question; propose the pipeline derive the family from the existing enum rather than adding a second stored field. | | clumping | yes, `clumping`, marked "invisible" | hard filter | Already there; this doc disagrees only with "invisible" — clumping changes maintenance pattern enough (scoop daily vs. dump and replace) that it is worth a visible chip, not just a backend filter. | | scented | no | hard filter (boolean) | Not in the add-on today. Two independent sources (OSU, ASPCA) treat scent as the single biggest cause of litter-box avoidance after texture, and it is on nearly every listing title/bullet ("unscented"/"lavender scent" etc.), so it is cheap to add and worth adding. | | multi_cat_formula | no | hard filter (boolean) | Titles reliably say "multi-cat." Real marketed sub-line (Arm & Hammer, Tidy Cats, others), not a generic household-size filter copied from elsewhere. | | texture_size | no | high_signal_markdown, candidate hard filter later | Pellet vs. fine grain is usually visible in the title or a photo-adjacent bullet, but I found no universal listing convention (some brands say "fine clumping," most plant pellets just show "pellets" in the name). Start it in the markdown line; promote to a filter if the scraper shows it reliably present. | | dust_level | no | high_signal_markdown | Front-of-bag marketing claim ("99% Dust-Free"), not a measured value anywhere I found (see Seed). Belongs in the compressed markdown line, not a structured column, until an independent test exists. | | odor_control_mechanism | no | high_signal_markdown | Named technology (baking soda, activated carbon) is reliably on brand spec pages, rarely in marketplace titles; a sentence, not a column. | | flushable_claim | no | hard filter (boolean) | Cheap, title-level claim ("flushable") when true. Should ship with the safety caveat (judgement J2) surfaced beside it in the UI, not as a separate stored field — this register treats that as a presentation decision, not a schema field. | Add-on fields this doc considers fine as-is: none dropped. The existing `litter_material` enum's `na`/`unspecified` values are sensible defaults for non-litter pet-care products that happen to share the add-on by accident of taxonomy; nothing here proposes removing them. ## Branches Each branch is a claim about how an expert narrows the choice, and each names the sentence in the register or in these notes it follows from. **Split 1, `litter_material` ("Clumping clay, crystal, or plant-based litter?")** - when `clumping clay`: then_ask `dust_level`, `texture_size`; makes_irrelevant `clumping`. - `clumping` irrelevant: the value chosen already fixes it — "are made from granulated bentonite clay, which clumps together when wet" (litter_material family description; Wikipedia, clumping-bentonite paragraph) means a buyer who has already said "clumping clay" has answered the clumping question by naming the family. - dust_level and texture_size then_ask: "No clay litter is completely dust-free" (Whisker, fetched excerpt) and "Clay litter is much more absorbent than sand and is manufactured into large grains or clumps of clay, making it less likely to be tracked" (Wikipedia, verbatim) — dust and granule size are the two things that vary most within the clay family and that a clay buyer specifically needs to weigh. - when `crystal (silica gel)`: makes_irrelevant `clumping`, `texture_size`. - `clumping` irrelevant: "No replenishing is necessary with silica gel (crystal) litter" (Wikipedia, verbatim), stated in contrast to clay/clumping litter needing topping up — crystal is defined by not clumping, so the question has no live answer left to give. - `texture_size` irrelevant: no source describes crystal litter in fine/medium/pellet terms the way clay and plant-based litter are described, which is an absence of evidence, not a positive claim that crystal ships as uniform beads or has no texture variation — softer footing than the `clumping` leg above, flagged as such rather than asserted with equal confidence. - when `plant-based`: then_ask `clumping`, `flushable_claim`. - `clumping` then_ask: "Guar gum was the principal clumping agent used in biodegradable cat litter" (Wikipedia, verbatim) stated alongside pine pellets and recycled newspaper as separate biodegradable sub-types — plant-based is the one family where clumping is not fixed by the family name; some plant litters clump (corn, wheat, walnut, with added gum) and some do not (pellets). - `flushable_claim` then_ask: "Other cat owners can be attracted to the biodegradable litters because of their flushability" (Wikipedia, verbatim) — the flushable marketing claim concentrates in this family, which is exactly why the safety caveat (judgement J2) needs to surface here rather than on the other two branches. **Split 3, `multi_cat_formula` ("One cat, or more than one?")** - Mapping rule (required because this split's facet is boolean but its question offers a plain-language choice): "more than one" maps to `multi_cat_formula = yes`; "one" maps to `no`. Stated in the split's own `rationale` field, not only here. - when `yes`: then_ask `odor_control_mechanism`. - Traced to: "Clump & Seal™ Multi-Cat, Complete Odor Sealing Clumping Litter" built on "ARM & HAMMER Baking Soda" (Arm & Hammer, fetched excerpt) — multi-cat formulas are specifically sold on a named odor-control technology, so a buyer who has just said "more than one cat" is the buyer for whom that technology name starts to matter. - No branch for `no`: a single-cat buyer is not told anything stops mattering — odor-control mechanism is still worth knowing, just less urgently — so nothing is forced irrelevant, consistent with "only write a branch where the answer really changes the path." Splits 2 (`scented`) and 4 (`texture_size`) carry no branches: both are independent reads that do not change what else gets asked, which the skill's own guidance says needs no branch ("independent questions need none"). ## Trimmed detail Facts and numbers cut from capped JSON fields or from facets I decided not to include, kept here so nothing sourced is lost. - Guar gum ≈ 69.8%, xanthan gum ≈ 73.1%, plantago gum ≈ 57.5% clumping rate (Vaughn et al. 2013, cited by Wikipedia) — single-paper number reached only via Wikipedia's citation, not fetched directly; kept qualitative ("plant-based litters need an added gum to clump at all, and gums vary in how well they work") in the trade-off entry instead of repeating the percentages. - "$2-billion industry consuming five billion pounds (2.3 Mt) of mined clay annually" (Wikipedia, verbatim) — used only as colour for the mining-impact trade-off's plain language, not as a sourced figure in a capped field. - A planned `self_cleaning_box_compatible` facet, dropped. I had one-sided brand evidence (Whisker markets its own clumping-clay litter as "developed to work flawlessly with Litter-Robot") but no source stating that crystal or pellet litter is *unsuitable* for self-cleaning boxes — the obvious Whisker help-center URL 404'd and the comparison-chart page, also fetched, did not mention litter compatibility. Asserting an incompatibility I only have the positive half of would have been exactly the kind of unsourced branch leg the checks.md history flags, so I left it out rather than guess. - "Some owners praise its absorbency because 4-5 pounds (1.8-2.3 kg) can absorb liquid and odor for up to 30 days for one healthy normal weight cat" (Wikipedia, verbatim, explicitly framed as an owner anecdote) and Arm & Hammer's "7 day odor-free home guaranteed" (fetched excerpt, marketing copy) — both considered for a numeric odor-duration facet and dropped; see Seed. - Litter depth (~2 inches, OSU and ASPCA both verbatim) and the "one box per cat plus one" rule — real, well-corroborated veterinary guidance, but about the box and household setup, not a property of the litter product itself, so kept here as corroboration for judgement J3 rather than turned into a facet. - Non-clumping clay (zeolite, diatomite, sepiolite) as a fourth `litter_material` value — considered and folded into the `clumping` facet instead, on the grounds that it is now a minority, legacy product and that keeping the split at three clean family values (clumping clay / crystal / plant-based) reads less like a history lesson and more like today's shelf. ## Corrections First run; no prior-run corrections to log. Lessons from `checks.md` applied up front rather than fixed after the fact: every number in `category.json` traces to a quote in this file labelled verbatim or fetched excerpt (none are search-excerpt-only); the two facets that looked like they might be "on the listing" by assumption (`dust_level`, `odor_control_mechanism`) were checked against what I actually fetched and both came back researched/claim-only, not listing-reliable in the structured sense; and the one branch I could not trace to a stated sentence (self-cleaning box compatibility) was dropped rather than kept with a weak justification. ## Corrections (2026-10-26) - Changed `dust_level.source` from `"listing"` to `"researched"` so the field matches what this file's own first-run Corrections paragraph already said about it. - Added `tracking_claim` facet (ordinal: standard / low-tracking claim), sourced to Fresh Step's own site nav, which names "Low Tracking" as a top-level feature filter and sells a "Clean Paws" low-tracking product line (fetched myself via curl, `freshstep_products.html`, `freshstep_lowtracking.html`), and to Whisker's existing "low dust, low tracking" claim already in this file. - Added `lightweight_formula` facet (boolean, listing), sourced to Arm & Hammer's own "Lightweight Formula" nav category (fetched myself via curl, `ah_lightweight.html`: "ARM & HAMMER™ Lightweight Cat Litter is 50% lighter than our regular litter") and to Fresh Step's parallel "Light Weight" product line (`freshstep_lightweight.html`: "35% lighter... for easier lifting and pouring"). - Reworded judgement J2 and `flushable_claim.why_experts_care`: the single Wikipedia sea-otter sentence supports cat feces reaching the ocean via urban runoff, not toilet-flushing specifically. Tried NOAA, Monterey Bay Aquarium, Oceana, Seattle Utilities and CDC pages for a second source tying the flushing pathway specifically to Toxoplasma; all 404'd, were empty JS shells, or blocked the fetch (consistent with this category's known fetch-blocking), so I reworded to the narrower claim rather than invent the stronger one. - Added vocabulary entries for "Proposition 65" and "Toxoplasma gondii," both used in facets/judgements without a definition. - Softened the crystal branch's `texture_size` justification above (Branches section) from an assertion that crystal "ships as uniform beads" to an explicit absence-of-evidence framing; I did not find a source stating crystal litter's bead size is uniform (tried Dr. Elsey's, PrettyLitter, Chewy within budget), so the branch itself is unchanged but its confidence is now honestly lower than the `clumping` leg next to it.