Extract Characters

Character extraction identifies all actors—humans, groups, organizations, animals, or abstract entities—in prose or narratives and resolves their aliases and pronouns into deduplicated entities. The Skill produces structured character profiles with canonical names, roles, motivations, and interaction maps.

Yes. Coreference resolution handles pronouns, nicknames, and alternate names to deduplicate references to the same character. The output lists each unique character once with all known referents consolidated.

Character extraction applies to narratives, reports, transcripts, and any prose or structured data where actors take action. It handles fiction, non-fiction, interviews, and documents with multiple human or non-human participants.

The Skill manages edge cases like unnamed narrators, plural groups, and collective actors by assigning descriptive identifiers and preserving their narrative role and interactions even when no proper name exists.

Character extraction outputs per-character blocks containing canonical name, aliases, role, motivations, and interaction summaries with other identified characters, formatted for downstream analysis or integration.

No special preprocessing is required. Provide raw prose, narrative text, or transcript data directly; the Skill handles standard text formats and resolves referents internally without external dependencies.