Padlet

Padlet MCP Server

Work with your Padlet boards from GitHub Copilot and any other AI agent that speaks the Model Context Protocol (MCP).

Padlet is a visual board for collecting and sharing ideas, used by teachers, students and teams. A board (a "padlet") holds posts with text, links, images, video and files, grouped into sections. This server lets an agent find, read, create and edit your padlets for you, in plain language.

• Hosted by Padlet. Nothing to install or run. The server is at https://mcp.padlet.com/ (streamable HTTP).
• No API key. You sign in with your own Padlet account in the browser the first time your client connects (OAuth).
• Acts as you. The agent can see and change only what you can see and change in Padlet.

This repository holds documentation only. The server itself is run by Padlet; its source code is not published here.

Contents

• Set up in VS Code with GitHub Copilot
• Copilot cloud agent (coding agent)
• Any other MCP client
• Signing in
• What you can ask for
• Tools
• Toolsets
• ChatGPT
• Limits and safety
• Support, privacy and terms

Set up in VS Code with GitHub Copilot

One click:

By hand: run MCP: Open User Configuration from the Command Palette (or create .vscode/mcp.json to share the setup with everyone working in a repository) and add:

{

"servers": { "padlet": { "type": "http", "url": "https://mcp.padlet.com/" } } }

Or from a terminal:

code --add-mcp '{"name":"padlet","type":"http","url":"https://mcp.padlet.com/"}'

Then open Copilot Chat in Agent mode. When the server starts, VS Code asks you to sign in to Padlet; see Signing in.

Copilot cloud agent (coding agent)

Not supported at the moment. GitHub's documentation says the Copilot cloud agent and Copilot code review "do not currently support remote MCP servers that leverage OAuth for authentication and authorization". Padlet's server signs you in with OAuth only and has no API-key option, so it can't be added there. Use Copilot Chat in VS Code instead.

Any other MCP client

Any client that supports remote MCP servers over streamable HTTP with OAuth can connect. Most clients take a config like this:

{

"mcpServers": { "padlet": { "type": "http", "url": "https://mcp.padlet.com/" } } }

The field names vary a little between clients (servers or mcpServers, http or streamable-http), so check your client's documentation. The URL is always https://mcp.padlet.com/.

Signing in

1. The first time your client connects, it registers itself with Padlet and opens a Padlet sign-in page in your browser.
2. Sign in with your Padlet account if you aren't already, and approve the request.
3. Go back to your client. You're connected.

Your client keeps the sign-in, so you won't be asked again each time. Every tool call runs as the signed-in user and follows Padlet's normal permissions: the agent can open the padlets you can open and change the ones you can change, nothing more.

What you can ask for

For teaching and classroom work:

• "Make a padlet called *Persuasive Writing: Week 3* with sections for Ideas, Drafts and Feedback."
• "Read my *Exit Tickets* padlet and summarise what students found hardest."
• "Comment on each post in the Drafts section with one strength and one next step."
• "Freeze my *Unit 2 Reflections* padlet so no one can post any more."

For teams and projects:

• "Find my board about the Q3 retro and group the posts by theme."
• "Add each item in TODO.md as a post on my *Sprint Planning* board."
• "Export the *Design Feedback* padlet as a PDF."
• "Change the *Launch Plan* padlet to a timeline layout with a dark wallpaper."

The agent replies with links to the padlets it worked on, so you can open them in Padlet.

Tools

With the plain URL https://mcp.padlet.com/ you get these 21 tools, whatever your library role:

The last five reach the rest of Padlet: padlet settings and moderation, collaborators, submission forms, breakout links, freezing, remaking, transferring, exporting, restoring deleted posts, whiteboard tools, automations, grading and grade sync to your LMS, Arcade activities, pinning and connecting posts, your account and the current time. If you are an owner, admin or teacher in a Padlet for Schools or Teams library, they also reach library administration.

The tool names in the toolsets below still work on the plain URL, so a client that saved them keeps working. The older tools these tools replaced, such as get_padlet_content, list_padlets and update_posts, no longer work anywhere. A call to one of them gets an error that names the tool to use instead.

Toolsets

To load named tools instead, add ?toolsets= with a comma-separated list. Some clients limit how many tools a model can see across all connected servers, so you can also load only part of Padlet:

https://mcp.padlet.com/?toolsets=core

https://mcp.padlet.com/?toolsets=core,board_admin

?toolsets=core,utility,board_admin,library gives 41 tools, or 65 if you are an owner, admin or teacher in a Padlet for Schools or Teams library. The library toolset appears only for those accounts. ?toolsets=core gives 17 tools (the 15 core tools plus the 2 upload tools). Unknown toolset names are ignored.

For a strictly read-only connection, use https://mcp.padlet.com/read-only. It serves 4 tools that only find and read padlets (get_current_user_info, find_padlets, get_padlet, export_padlet) and can't create or change anything.

All tool names by toolset

core: get_current_user_info, find_padlets, get_padlet, export_padlet, create_padlet, update_padlet_heading, save_posts, save_sections, save_comments, delete_padlet_items, analyze_post_attachments, restore_posts_and_sections, pin_post, create_post_connection, delete_post_connection. Upload tools served with core: start_attachment_upload, complete_attachment_upload.

utility: get_time

board_admin: update_padlet_appearance, update_padlet_layout, update_padlet_post_fields, update_padlet_engagement, get_padlet_content_settings, update_padlet_content_settings, update_padlet_sharing, invite_email_as_padlet_collaborator, update_delete_padlet_collaborator_user, update_submission_form_settings, update_breakout_settings, freeze_padlet, create_automation, delete_automations, get_padlet_automations, remake_padlet, transfer_padlet, update_auto_remake_settings, create_padlet_arcade_activity, grade_posts, sync_grades_to_lms, get_whiteboard_tool_settings, update_whiteboard_tool_settings

library: list_libraries, get_library_info, get_library_invite_links, update_library_invite_links, send_library_email_invites, remove_library_email_invites, bulk_invite_library_members, get_library_onboarding_flow, complete_library_onboarding_step, get_library_oauth_settings, update_library_oauth_settings, get_library_saml_settings, create_library_saml_connection, update_library_saml_connection, validate_saml_metadata_url, get_library_approved_domains, request_library_domain_approval, remove_library_approved_domains, get_library_permissions, update_library_permission, get_library_content_safety, update_library_content_safety, get_library_attachment_type_settings, update_library_attachment_type

ChatGPT

Padlet's app in ChatGPT connects to this same server, but lists one tool for each operation. It has no action tools. It lists the first 16 tools above, plus each toolset tool that does a job none of them covers, such as freeze_padlet, remake_padlet and the library tools. A new Padlet tool appears in ChatGPT after OpenAI approves it.

Limits and safety

• Everything runs as you. There is no shared service account. Library tools still check your role in that library on every call.
• Changes you can undo, and changes you can't. Deleted posts and sections can be restored with restore_posts_and_sections. A tool is marked destructive when it overwrites what a person wrote, deletes something, or changes a whole library in a way that can't be put back exactly. Clients that honour MCP tool annotations can ask you before running those tools. On the plain URL they are update_padlet_heading, update_padlet_sharing, save_posts, save_sections, save_comments, delete_padlet_items, complete_attachment_upload and run_padlet_destructive_action, which runs every destructive action.
• Sandbox (whiteboard) padlets are only partly supported; the other padlet formats are fully supported.

Support, privacy and terms

• Padlet: https://padlet.com
• Help centre: https://padlet.help
• Contact: hello@padlet.com
• Privacy Policy: <https://