Tencent Docs

Problem

Tencent Docs MCP handles online document creation, reading, and editing rather than local file management. It routes reports, notes, articles, smart sheets, calculation sheets, slides, mind maps, flowcharts, Word files, and forms to the right tools, reducing manual rework between Markdown, spreadsheets, decks, and image recognition.

How it works

• Format routing: new documents prefer smartcanvas with create_smartcanvas_by_mdx, which accepts MDX and common Markdown; structured data uses smartsheet, filtering and statistics use sheet, presentations use slide, and knowledge graphs or processes use mind and flowchart.
• Read/write paths: content retrieval uses get_content, file renaming, moving, deletion, copying, and permissions use manage.*; web clipping uses scrape_url and scrape_progress; OCR uses ocr.extract, ocr.toword, and ocr.toexcel.
• Uploads and async tasks: local file persistence usually follows a unified import and progress polling path, instead of recreating existing content with create tools; async tasks such as slide generation and web clipping should be polled in sub-sessions so the main session stays responsive.
• Error handling: auth, VIP, credit, and parameter errors need separate checks; tool parameters follow the MCP schema, and unsupported capabilities trigger report_unsupported_feature.

Boundaries

This skill depends on the Tencent Docs online service and local authorization. Some create_* tools support parent_id for placing files in a directory, but create_smartcanvas_by_mdx does not. A space node's node_id is also the document file_id, and node deletion can carry recursive-delete risk. OCR calls differ between local images and public images, and parameter conflicts should follow the actual tool schema.