Pokemon Team Builder
Choose validated species and forms, then configure ability, item, nature, Tera Type, moves, EVs, IVs, nickname, and level. Your draft stays on this device.
How to read the tool — Start with facts, then add context
The team you enter is treated as the evidence source: selected Pokémon, moves, items, abilities, EVs, IVs, nature, Tera type, and format context where supported. PokeTeamOn does not invent an opponent set, infer your battle plan, or turn incomplete data into a hidden score.
Fact → Context → Explanation → Optional repair. Build the team first, inspect what is actually present, test one concern, and change something only when the evidence gives you a reason.
Format context
Choose the battle context this team is being prepared for. The selection stays on this device and does not change your saved six members.
Import / Export Pokémon Showdown text
Export your current team or paste Pokémon Showdown text to import it. This free workflow stays local and keeps the selected format context.
Team legality check
Checks local set structure, Gen 9 move compatibility, and only format rules that PokeTeamOn can enforce conservatively from pinned local sources.
This is not a full format legality guarantee.
Team structure analysis
Summarizes types, selected move coverage, STAB presence, repeated weaknesses, and defensive resistances from pinned local data.
No team score or matchup prediction is produced.
Probe an opponent
Compare your selected team and moves with one opponent species or form using Gen 9 typing only.
Ability, item, Tera, damage, and speed effects are not modeled here.
Analyze team structure
Turn the current team into deterministic set-level and team-level structural evidence.
How to interpret Analyzer
What was detected: Analyzer describes structural facts from the team you entered, such as selected moves, offensive profile, speed distribution, priority access, and repeated typing exposure where the supported data allows it.
Why this matters: a finding is evidence for a diagnosis, not the diagnosis itself. Repeated Ground weakness, for example, is worth attention only in the context of the team's answers, role plan, and the problem you are trying to solve.
Evidence boundary: Analyzer does not simulate a hidden opponent, estimate a win probability, or decide which signal is most important for you.
Next step: pick one suspicious finding, read its explanation, then stress-test that exact concern instead of changing several team members at once.
No team score, damage simulation, opponent scenario, stress test, or automatic repair.
Pressure-test explicit assumptions
Select only the checks you want. Thresholds and target types are never filled from hidden meta defaults.
How to interpret Stress Test
Declared condition: every stress test begins with something you explicitly chose — for example a Speed floor, incoming type pressure, a coverage target, or emergency priority access.
What was assumed: the threshold or target is your test condition. It is not a claim that the condition is universally important in every matchup.
Read the result narrowly: pass, mixed, or fail evidence describes how the current team meets that declared condition. It is not a prediction of an unknown opponent or a forecast of the battle result.
Next step: keep the condition fixed when comparing before and after. That makes a later repair testable instead of turning it into a vague improvement claim.
Choose at least one explicit check. No team score, win probability, damage simulation, or automatic repair is produced.
Minimal-change repair preview
Generate deterministic move repairs for one explicit goal. Nothing is applied until you select a candidate, preview it, and confirm the local apply action.
How to interpret Repair
Diagnosed target first: Repair starts from one explicit problem you selected. It does not search for a mysterious "best team" or silently choose what matters most.
Minimal-change boundary: candidates are limited to the mutable slot and supported goal you provided. Candidate order is deterministic for reproducibility; it is not a ranking and is not an automatic recommendation.
Preview before apply: inspect what changes and what stays untouched before you confirm. The server does not apply or persist the repair for you.
Next step: after a local apply, rerun the same Analyzer finding or Stress Test condition. The useful question is whether that specific evidence changed, not whether the tool can promise a win.
Choose one explicit repair goal and mutable move slot. Candidate order is deterministic, not a ranking.
The server never applies or persists repairs. Local team storage changes only after you click Apply selected repair and confirm.
EXPLAIN — Evidence behind the latest result
Run Analyzer, Stress Test, or Repair first. Explain uses only the latest captured deterministic result and does not rerun it.
Only facts from the captured Basic Analysis are described.
Only factual evidence from the captured result is described.