Official Document Word Assistant

Problem It Solves

When working with DOCX official documents, the hard part is not producing text. It is checking whether agency names, document numbers, recipients, version blocks, fonts, tables, headers, footers, and redaction artifacts remain consistent. Manual page review can miss details, while conversion can break the original layout. gongwen-word-mini turns drafting, revision, formatting, and validation into a state machine: run doctor, build a request JSON, run inspect to check document_class, official_intent, strictness, and direction, then use build for drafts or prepare for formal delivery.

Core Capabilities and Workflow

• Conservative facts: missing fields must be written as [pending: field]; strict mode blocks formal delivery when placeholders exist.
• Original-file protection: DOCX edits must save to a copy, and the tool rejects identical input/output paths.
• Machine checks plus visual review: prepare runs generation, validation, page-by-page PNG preview, and a visual review template; until a human completes review, delivery_ready remains false.
• Editing and redaction: documents with images or tables use an OOXML fidelity path, while ordinary documents use python-docx; redaction mappings must be supplied by the user and are not echoed in reports.
• Boundary disclosure: seals, strict version blocks, joint issuance, multiple agency signatures, or institution-specific letterhead require the organization’s approved template and visual review.

It is better understood as a gated document engineering workflow than a plain text polishing tool. It suits formal materials that must move from “readable” to “deliverable”; it is not a tool for inventing agency metadata, adding red-letter elements arbitrarily, or relying on rendering alone to claim strict formatting compliance.