document-editor

Creates and edits long-form Markdown documents in a rich text editor with streaming updates.

1.2k|166|Updated Feb 7, 2026
One-click install
npx skills add https://github.com/vellum-ai/vellum-assistant --skill document-editor
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: document-editor
Source: https://github.com/vellum-ai/vellum-assistant/tree/main/assistant/src/config/bundled-skills/document-editor
Command: npx skills add https://github.com/vellum-ai/vellum-assistant --skill document-editor

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Long-form writing like articles, blog posts, essays, and reports gets lost or mangled when dumped into chat replies. This Skill routes prose into a dedicated rich text editor where content can be streamed, reviewed, edited, and exported instead of cluttering the conversation.

Core Features & Use Cases

  • Streamed document creation: Create a document with a title and stream Markdown content in chunks via append updates, so users see text appear in real time.
  • Targeted editing: Use find-and-replace with literal or regex patterns (including backreferences) to fix typos, rename terms, or reformat content without rewriting the whole document.
  • Comment workflow: Read inline user comments, edit the document to address feedback, reply for clarification, and resolve threads.
  • Use Case: A user asks for a 1,500-word blog post about remote work trends. The assistant creates a document, streams sections into the editor, then later fixes a recurring typo and reformat dates using regex replacement when the user requests revisions.

Quick Start

Ask the assistant to write a blog post about a topic of your choice and it will draft it in the document editor.

Frequently Asked Questions about document-editor

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I write a blog post or article with an AI assistant?

Ask the assistant to write or draft your article, and it creates a document in the rich text editor rather than replying in chat. Content is streamed in chunks so you can watch it appear and edit it directly afterward.

How do I edit an existing document without rewriting everything?

Targeted edits use document_find to preview matches and document_replace_text to apply literal or regex replacements, including backreferences. Full-document replace mode is reserved only for complete rewrites where most content changes.

Can I find a document from a previous conversation?

Yes, document_list with a title query searches documents, and guardian or local users can search across past conversations and sessions. Once found, document_open re-surfaces the editor panel on your current device.

What happens if a document update fails mid-stream?

Resend the failed document_update call with the chunk's Markdown in the content field, omitting surface_id to target the current document. Do not call document_create again, since that produces a duplicate document.

When should I use the document editor instead of an app or chat reply?

Use the document editor for prose like articles, essays, and reports that users will review or export. Interactive content with state or data belongs in an app, and one-or-two-sentence answers belong in chat.