Transparent by design
How FishyMag decides a verdict
This dataset can harm animals if it is wrong. That single fact shapes every decision below.
Living system · illustrativeThe two rules
Both are enforced in code and checked on every build, not left to editorial care.
1. Ingredient evidence may raise risk. It may never establish safety.
Copper is the best-known invertebrate killer in the hobby, so it is tempting to clear anything copper-free. But Fritz Algae Clean Out contains no copper and is still explicitly unsafe for shrimp and snails. A checker built on ingredient inference would have issued a green tick on that product. So the inference runs one way only: a known-harmful ingredient raises a warning, and a clean ingredient list proves nothing at all.
2. Absence of evidence is Unknown. It is never Safe.
Most labels simply never mention shrimp. Silence is not a safety claim, so a missing statement produces Unknown and an instruction to ask the manufacturer. A positive verdict requires the manufacturer to have said so, in words we can quote.
The five verdicts
A binary safe/unsafe flag would force every undocumented case into one of two buckets, and both would be lies.
- Manufacturer states compatible
- The manufacturer states it can be used with this group. The strongest positive state, and the only one that requires a first-party source.
- Manufacturer advises against
- The manufacturer states it should not be used. The strongest negative state.
- Caution
- Permitted but qualified — a reduced dose, a warning that reactions vary, or an instruction to remove livestock first.
- No evidence found
- No statement was found. Treat it as a no until you have checked, not as a yes.
- Sources disagree
- Credible sources disagree, including cases where a manufacturer contradicts itself on one page. Both are shown so you can judge.
Source authority
Where two sources disagree, the lower number wins — and the disagreement is disclosed rather than hidden.
- 1manufacturer label
- 2manufacturer page
- 3manufacturer faq
- 4sds
- 5ingredient inference
- 6retailer
- 7community
The bottom three — ingredient inference, retailers and community reports — can raise a concern but can never, on their own, justify a manufacturer states compatible verdict. Retailers have an incentive to sell, and forum consensus is not testing.
What this site does not do
- No diagnosis. FishyMag will not tell you what your fish has, or which medication to reach for. It answers one question: can this treatment be used with what is already in your tank, according to the people who make it.
- No claims of our own about toxicity. “The manufacturer advises removing invertebrates” is a claim we can defend. “This kills shrimp” is not, so we do not make it.
- No paid verdicts. No manufacturer can pay to change a rating, and none has been consulted on any verdict here.
Formulations, not products
Every record describes a specific formulation in a specific market. The same brand name can be reformulated over time or sold with a different composition elsewhere, and a verdict attached to a name rather than a formulation eventually describes a bottle nobody can buy. When a manufacturer changes a label materially, the affected verdicts are marked for review rather than left showing an old result.
Current coverage
- Formulations
- 8
- Manufacturers
- 3
- Ingredients
- 3
- Verdicts published
- 41
- Backed by a first-party source
- 27 of 41
Dataset last checked August 16, 2026. Coverage is deliberately narrow: a small dataset where every verdict is traceable is worth more than a large one where some are guesses.
Corrections
If a manufacturer has changed a label, or you believe a verdict misrepresents a source, that correction matters more here than on an ordinary site. Errors in this dataset cost people their livestock.