Entities

Entities are the main things Arc tracks in a manuscript.

An entity can be:

• a character;
• a place;
• an object;
• an organization;
• an event;
• a concept, rule, or recurring motif.

Canonical identity

Each important entity should have a stable identity. That identity can include many surface names.

For example, one character might appear as:

• Mara;
• Captain Mara;
• the captain;
• Mara of the River Gate.

Arc should try to connect these references without pretending uncertainty is certainty. When the source is ambiguous, the item should remain reviewable.

Entity fields

An entity may include:

• canonical name;
• aliases;
• type;
• description;
• source evidence;
• related entities;
• review state;
• notes or author decisions.

The exact fields may change as the product evolves, but the principle should not: entity data must remain inspectable and traceable.

Entity quality

A useful entity is not merely extracted. It is clear enough that the author can answer:

• What is this?
• Where did Arc find it?
• Is this the same as another entity?
• What relationships matter?
• Is this accepted canon or still uncertain?

Common problems

Entity review often involves:

• duplicate characters;
• aliases not connected to the canonical entity;
• places mistaken for organizations;
• events mistaken for objects;
• pronouns or titles attached to the wrong person;
• concepts that should remain notes rather than canon entities.

These are normal first-pass issues. The review workflow exists because extraction should not silently become truth.