shard-document

Document sharding breaks large documents (over 5,000 words) into logical, navigable pieces while preserving cross-references and dependencies. This Skill splits PRDs, specs, architecture docs, and API references using multiple strategies—logical, size-based, hierarchical, or semantic—and generates an index shard plus metadata for each piece.

Sharding preserves all cross-references and dependencies between document pieces. The Skill validates internal links and generates rich per-shard metadata so relationships remain intact across shards, plus produces a navigation index to guide readers through the fragmented content.

Yes. The Skill supports multiple sharding strategies: logical (by section structure), size-based (by word count), hierarchical (by document outline), and semantic (by meaning). You select the strategy that fits your document type and navigation needs.

Sharding produces a directory with an index shard (navigation guide), individual shard files for each document section, and per-shard metadata. The index and metadata preserve document structure and enable readers to navigate between related pieces.

Yes. Sharding is designed for long-form technical documents including API references. It breaks them into navigable sections while maintaining endpoint relationships and cross-references, making large reference docs easier to explore and maintain.

Sharding works best for documents exceeding 5,000 words where navigation becomes difficult—PRDs, detailed specs, architecture docs, and API references. It's unnecessary for short documents but improves usability and maintainability for lengthy technical content.