Using boards, documents, and lists in the Eden MCP

This article assumes you already have Eden connected through the MCP. If you don't, start with connecting AI assistants.

Why attach a board (instead of asking each time)

The contents of your "Hooks" board are mostly the same in two questions you ask back to back. If you have to remind the assistant about them on every question, the conversation gets clunky fast.

Attaching solves that. You tell the assistant once at the start of the chat, "here's my Hooks board." The assistant reads it, holds onto it for the whole conversation, and references it on every reply. No re-pasting. No "remember what's on the board." It just does.

This is what Eden makes available to attach.

The three things you can attach

A board

The full contents of a board, rendered as one long document the assistant can read top to bottom. Every saved card with its URL, every sticky note, every text block, every label, and every section divider.

Attach a board when:

• You're drafting from a swipe file ("write a tweet in the style of the hooks on this board").
• You'll ask several follow-ups about the same material and don't want to keep telling the assistant where to look.

The attach menu lists every board you can see across every workspace you belong to, with the most recent ones first.

A document

A single document (Eden's word for a markdown note). The assistant reads the full body, exactly as you wrote it.

Attach a document when:

• You want the assistant to read, edit, or build on a specific document you already have.
• You need the whole thing, not a search snippet.

Documents don't show up in the attach menu directly (a productive workspace has too many of them to list cleanly). The reliable way: ask the assistant to find one first ("find my document about Q3 planning"). It'll surface the right document and you can attach it from there.

A creator list

The creators in one of your curated lists, with handle, platform, follower count, and bio.

Attach a list when:

• You want the assistant to focus on a specific group of creators you've grouped together ("compare the writing styles in my 'Newsletter inspirations' list").
• You want to brief the assistant on the people relevant to one project.

Only lists you built yourself show up in the attach menu. Auto-generated category lists are filtered out so the menu stays clean.

How to attach in each assistant

Claude (Desktop or web)

Claude has the smoothest attach flow of the three.

1. 01

   Open the attach menu in the chat composer

   Click the + (or paperclip) icon in the composer. You'll see options for files, screenshots, and "From integrations."

2. 02

   Pick Eden, then the type of thing you want to attach

   The Eden submenu shows Boards, Documents, and Creator lists. Pick the type.

3. 03

   Choose the specific item

   Claude lists everything available with clear names (e.g. "Hooks board, Personal Workspace"). Pick the one you want.

4. 04

   A chip shows what's attached

   You'll see a small chip in the composer showing what you attached. Every reply in this chat will have it available. Add more chips for more attachments (e.g. two boards to compare).

Cursor

Cursor's attach menu is on the @ key.

1. 01

   Type @ in the Composer

   A picker opens with files, symbols, docs, and anything from your connected MCP servers.

2. 02

   Pick what you want to attach from Eden

   You'll see Eden's boards, documents, and lists in the menu.

3. 03

   Choose the specific item

   Same as Claude. Pick the one you want; it attaches to the message.

In Cursor, attachments apply to the message you're writing, not automatically to the whole conversation. If you want a board to apply to every message, attach it each time, or use a Cursor rule to keep it persistent. We're working on a smoother flow for this.

ChatGPT

ChatGPT's attach experience for MCP attachments is the least polished of the three today. There's no dedicated picker in the composer yet. The reliable way:

1. 01

   Ask the assistant what you can attach

   "What Eden boards and lists can you read?" ChatGPT calls Eden, comes back with the list, and shows you what's available.

2. 02

   Ask it to use a specific one

   "Read my Hooks board, I'll ask you questions about it." The assistant reads the item and holds onto it for the rest of the conversation.

This isn't as crisp as Claude's attach menu, but it works. If OpenAI ships a dedicated picker for connector attachments, ChatGPT will start using it automatically.

A few specific workflows worth knowing

"Use this board as the source for the next ten questions"

Attach the board. Now follow-ups like "which of these three hooks would land best?" get answered from the board's contents without you re-pasting anything.

"Focus on this specific group of creators"

Attach a list. Now "what are the writing styles in this group?" or "which creator here would I learn the most from?" gets answered from the list, not from your whole following.

"Compare what's on my Hooks board to my Headlines board"

Attach both boards. Ask the comparison. You get a side-by-side without the assistant having to look either one up mid-reply.

What attachments do NOT do

A few things worth being clear on:

• Attaching is read-only — saving is a separate step. Attaching a board, document, or list just feeds it to the assistant as context to read; the attachment itself never changes. Saving back to Eden (a new card, a note edit, a scheduled post) is a separate write action the assistant takes with its tools, and only when the connection has write access — see Reading vs. writing. (Following creators and editing lists still aren't available through any connection.)
• No live updates. When you attach something, the assistant reads it as it was at that moment. If you change the board contents partway through a chat, the assistant doesn't see the change until you attach it again (or start a new chat).
• No "attach my whole workspace." You attach one specific board, document, or list at a time. The unit is one item, not "everything I have."
• No metrics or rankings. Attachments are for content (your saved cards and documents). For things like "top outliers from this creator" or "how did my posts do," the assistant uses its research and analytics tools, not an attachment.

Where to go next

• To run your Custom AI as /eden- slash commands, see using Custom AI as slash commands.
• To have the assistant adopt a full Custom AI — instructions plus its knowledge base — see using your Custom AIs through the MCP.