Tencent Docs MCP

Problem addressed

When an agent must move chat output, structured data, web pages, or local files into Tencent Docs, the task is not just document creation. It requires choosing the right doc_type, reading existing content, using the correct import, clipping, or OCR path, and diagnosing permission, parameter, or async task failures. This skill turns those choices into an execution guide.

How it works

• Default document path: new documents prefer smartcanvas, using create_smartcanvas_by_mdx with MDX; Markdown can be passed directly without format conversion.
• Document coverage: supports sheet, smartsheet, slide, mind, flowchart, form, doc, and smartpage, with scenario routing to the relevant reference docs.
• Common workflows: search with manage.search_file, then read with get_content; upload local files via import_file.sh, manage.async_import, and manage.import_progress; clip URLs with scrape_url and scrape_progress; process images with ocr.* or local scripts.
• Calls and debugging: use the schema from mcporter list tencent-docs; responses include error and trace_id; authentication, VIP, credit, and parameter errors can be handled by error code.

Boundaries and notes

Useful for agent development that calls Tencent Docs APIs repeatedly. First use requires local authorization; some create tools do not accept parent_id; delete_space_node with all deletes recursively; async tasks are best polled in a child session; unsupported capabilities should be reported silently.