/**
 * Document Chunker — Split markdown files into searchable chunks
 *
 * Strategy:
 * 1. Split on H2 headers first (natural semantic boundaries)
 * 2. If a section exceeds target size, split on paragraph boundaries
 * 3. Each chunk carries metadata: file path, title, category, chunk index
 *
 * Target chunk size: 500-800 tokens (~2000-3200 chars)
 */
export interface DocChunk {
    file_path: string;
    chunk_index: number;
    title: string;
    section_title: string;
    category: string;
    content: string;
    file_hash: string;
}
export interface DocFile {
    absolute_path: string;
    relative_path: string;
    content: string;
    hash: string;
}
/**
 * Chunk a single markdown file into searchable segments
 */
export declare function chunkDocument(doc: DocFile): DocChunk[];
/**
 * Scan a directory for markdown files
 */
export declare function scanDirectory(dirPath: string, options?: {
    exclude?: string[];
}): DocFile[];
/**
 * Chunk all markdown files in a directory
 */
export declare function chunkDirectory(dirPath: string, options?: {
    exclude?: string[];
}): {
    files: DocFile[];
    chunks: DocChunk[];
};
//# sourceMappingURL=doc-chunker.d.ts.map