Neon Abyss 2 Characters & Agents

Why the table does not guess missing fields

Searchers need accurate starts and unlocks, not confident placeholders. Character kits are especially vulnerable to update drift because a familiar name can remain while an ability, weapon, resource, or unlock requirement changes. TBA is actionable: it tells an editor exactly what must be captured from the live game.

How to verify one roster row

Open the current character-select screen and profile. Record the displayed name, ability text, starting weapon, starting health or resources, locked or unlocked state, and any stated requirement. Include the game version and date. Then start a run to confirm that the displayed loadout appears in play.

With the roster boundary clear, selection becomes a question of loop and execution rather than name recognition.

Comfort is a real build advantage

An Agent whose movement and starting attack you control consistently reaches more drops with more health. That extra route depth can matter more than a theoretically stronger ability you activate late or use unsafely. Test the first room instead of assuming difficulty from the profile wording.

A starting identity is not a permanent class

The opening kit determines the first decisions, not the entire run. A later weapon, Curio chain, Hatchmon, or Faith choice can shift the best range and responsibility. Keep the useful parts of the Agent kit, but do not reject a proven pivot merely to preserve the original label.

A good choice still depends on understanding exactly what the Agent brings into that first room.

Verify the screen and the run

The profile explains intended behavior; the first room shows practical behavior. Confirm both. A tooltip may omit cadence, movement commitment, projectile path, or how a special resource refills. Report those observations separately from the written rule.

Do not convert one video into a full roster table

The selected gameplay video shows a player discussing opened characters and favoring an extra-heart opening, but it does not display enough readable evidence to assign every name a start. It supports the decision concept, not ten specific claims.

Unlock requirements need the same per-row evidence discipline.

Use a two-step unlock check

Before completion, capture the locked profile and requirement. After completion, return to character select and capture the selectable state. If the Agent remains locked, record missing prerequisites rather than rewriting the requirement from memory.

Keep R-7 and Tino explicitly unverified

R-7 and Tino belong in the research roster, but their exact version 1.0 availability and unlock paths remain TBA here. This is stricter than claiming they are absent or automatically unlocked; either statement would require current evidence.

Once an Agent is available, its role should be described through actions the build performs rather than a permanent class badge.

Assign roles after seeing the starts

A party should inspect each Agent and opening weapon before deciding who covers range, contact, crowds, or bosses. Assigning “tank” by appearance or “support” by name invents mechanics. Observable reach, safety, trigger frequency, and resources are better inputs.

Reassign after major pickups

If a new weapon gives one player safer boss uptime or a Curio chain gives another crowd control, move the responsibility. The Agent identity still matters, but the complete current build decides execution.

Players who want to avoid habit entirely can use the version 1.0 random-selection feature as a deliberate practice tool.

Random does not remove the reading step

After the result appears, pause and inspect the ability and start. State the opening loop and weakness before entering. Random selection should create variety, not excuse a blind first room.

Compare outcomes through profiles

When trying random Agents, use profile records and your own dated notes to separate Agent comfort from one lucky item run. Boss kills and highest Fate can suggest where more testing is needed, but they are not a universal strength ranking.

Those records are part of the expanded Agent Profiles introduced in 1.0.

Use profiles for personal planning, not global tiers

A high Fate record can reflect player familiarity, a fortunate build, co-op context, or an older balance version. It is useful for choosing the next personal challenge but insufficient for claiming one Agent outranks another.

Build a versioned character record

For each verified Agent, save profile wording, selection-screen start, first-room behavior, unlock evidence, mode, version, and date. That small record supports an accurate wiki row and makes later patch comparisons straightforward.

The FAQ closes with short answers for roster, unlock, and selection searches.