What is Frame.io MCP?
Use your MCP-compatible AI agent — Claude, ChatGPT, and many more — to connect to our Frame.io MCP server and control Frame.io directly through natural language. This feature is available to all plans, starting with a public beta.
MCP (Model Context Protocol) is an open standard for connecting AI clients to external systems. The Frame.io MCP Server exposes a set of tools — organized operations your AI agent can call — so that when you make a request, the agent picks the right tool, calls the server, and uses the response to complete your task.
Think of it as a direct line between your AI assistant and your Frame.io account. Instead of switching tabs, searching manually, or uploading files through the web app, you instruct your AI agent in plain language and it handles the action. Or ask it any question about your account to find all relevant material.
Example actions include creating folders, uploading files, managing share links, pulling audit logs, managing access, and much more.
The Frame.io MCP server is hosted at https://mcp.frame.io/v1/mcp.
Note: This feature is independent from the Frame.io AI Assistant. MCP is an external user-controlled integration using LLMs, while the Frame.io AI Assistant is an in-product Frame-controlled chat feature.
Actions with Frame.io MCP
Here are some examples of what you can ask once connected:
"Show me all the projects in my main workspace."
"Find any files with "rough", "draft", or "WIP" in the name across my account."
"Upload my asset to Frame.io, generate a share link, and request a review from the following people."
"What is the current storage utilization per workspace and per project?"
"Generate a report of all approved files over the last two weeks, including who commented and who approved each file."
"List everyone that has logged into my Frame.io account in the last 30 days."
Full list of available actions
Below is a list of all available actions across multiple categories in Frame.io. Essentially, everything a Frame.io content admin can do in the UI, an AI agent can do through natural language.
Category | Key Capabilities |
Accounts & Workspaces | List, create, rename, delete; full audit log access |
Projects | List, create, archive, update, delete |
Files | Upload from local or URL, move, copy, rename, delete, check upload status |
Folders | Create, rename, move, copy, list contents, delete |
Version Stacks | Create, reorder, move, copy, list versions |
Search | Search assets across name and content, account-wide |
Share Links | Create public or secure links, manage reviewer access, set expiry/download settings |
Comments | Create, edit, timestamp-anchor, mark complete, delete; attach files to comments |
Metadata | Define fields, set values in bulk across multiple files |
Members + Access | Manage roles at account, workspace, project, and folder level |
Audit Logs | Full history of who did what and when (read-only) |
Recently Deleted | List and restore deleted items |
Note: There are a select few actions the MCP server cannot perform at this time during the beta. We will update this list if these features become available.
Read or understand visual content — it reads file names and folder structure, not what's inside a video or image.
Deliver Frame notifications proactively to an AI agent.
Integrate with Frame's automation features, like webhooks or custom Actions.
Support Legacy Frame.io accounts.
Switch between multiple Frame accounts in a single AI session (unless they're connected to a single login).
Before you connect
Requirements
You must have these requirements in order to use Frame.io MCP:
An active Frame.io account on V4
An Adobe ID associated with a Frame User
An AI client listed in the Connecting your AI client section
Frame Managed vs. Adobe Managed Accounts
Frame.io has two types of user accounts, and knowing yours matters for authentication.
Frame Managed Accounts (FMA): Created and managed directly within Frame.io. Sign in with your Adobe credentials at the login prompt and you're connected.
Adobe Managed Accounts (AMA): Administered through the Adobe Admin Console, typically for enterprise teams. The same Adobe identity can belong to more than one organization — whichever organization you authenticate into determines which Frame.io account the MCP Server can access.
Note: If you're on an AMA and see empty results or permission errors after connecting, you may have authenticated into the wrong organization. Clear your Adobe browser session and reconnect to select the correct profile.
Connecting your AI client
Authentication uses OAuth 2.0 through Adobe IMS and runs once on first connection. After that, token refresh is handled automatically.
Claude (Desktop or web)
Step 1: Open Claude Desktop. From the menu bar, open Customize and select Connectors in the left sidebar. Click + in the top-right corner.
Step 2: In the Add custom connector dialog, enter:
Name: Frame.io
Click Add.
Step 3: Frame.io appears in the Not connected section. Click Connect. You'll be taken to Adobe IMS — sign in with the Adobe ID associated with your Frame.io account. After authorizing, you're returned to Claude Desktop.
Step 4: Once authenticated, Frame.io moves to the Web section. The right panel lists tools split into Interactive tools and Read-only tools. Set Read-only tools to Always allow. Leave all other tools on Needs approval until you've reviewed tool permissions below.
To confirm the connection is working, ask Claude "List my Frame.io accounts."
ChatGPT (Desktop or Web)
Step 1: In ChatGPT, go to Customize → Plugins in the sidebar.
Step 2: Click Add in the top right, then Create custom MCP server. Fill in:
Name: Frame.io
Description: Connect to your Frame.io account
MCP server URL: https://mcp.frame.io/v1/mcp
Authentication: OAuth (do not enter a bearer token)
Step 3: Check "I understand and want to continue", then click Create as plugin.
Step 4: On the Connect Frame.io screen, click Continue to Frame.io.
Step 5: You’ll be redirected to Adobe’s sign-in page. Sign in with the Adobe ID associated with your Frame.io account.
VS Code
Step 1: Open the Command Palette (Cmd+Shift+P on macOS / Ctrl+Shift+P on Windows/Linux) and run MCP: Add Server.
Step 2: When prompted, select HTTP (HTTP or Server-Sent Events), then enter:
Name: frameio-mcp
Scope: Global (available across all workspaces)
Step 3: A browser window opens to Adobe IMS. Sign in and authorize access.
Step 4: Run Chat: Manage Tool Approval from the Command Palette to set per-tool permissions. Open the Chat view (Ctrl+Alt+I / ⌃⌘I) and switch to Agent mode. Frame.io tools are now available.
Cursor
Note: Requires Cursor 0.45 or later.
Step 1: Create ~/.cursor/mcp.json with the following:
{ "mcpServers": { "frameio-mcp": { "url": "https://mcp.frame.io/v1/mcp" } } }
Step 2: Open Cursor Settings (Cmd+, / Ctrl+,) and navigate to Features → MCP. frameio-mcp will appear in the server list.
Step 3: Click Enable next to frameio-mcp. Cursor opens a browser window to Adobe IMS. Sign in and authorize access.
Step 4: Open a Composer window. Frame.io tools are now available. To confirm the connection, ask: "List my Frame.io accounts."
Example Workflows
Once you're connected, here are four examples of how teams can use Frame.io MCP server to its greatest potential.
Example 1: 📋 Audit & Reporting
Generate a report of all share links created in the last 30 days, including who accessed each link and whether it was public or secure
List everyone that has logged into my account in the last two weeks, with date and time
Show me all approved files from the Fall Campaign project, including who approved each one
Example 2: 📤 Share for Review
Upload my asset to Frame.io, generate a share link, and request a review from [names/emails]
Get all of the comments on my share
Upload the new version of my asset and update the same share link
Example 3: 🔍 Browse & Discover
Show me all active projects in my main workspace
Find any files with "rough", "draft", or "wip" in the name across my account
What's the upload and processing status of everything in the Final Deliverables folder of the Q3 Campaign project?
Example 4: 🗂️ Organize
Create a folder structure for the new campaign project
Move all files tagged "approved" into the Final Delivery folder
Stack these two files as versions of the same asset
Troubleshooting
Here are the most common troubleshooting occurrences and solution you may come across.
Issue | Fix |
OAuth loop or authentication failure | Sign in with the Adobe ID linked to your Frame.io account. If you have multiple Adobe accounts, confirm you're using the correct one. Removing and re-adding the server starts a fresh OAuth session. |
AMA: authenticated but seeing wrong account / empty results | Clear your browser cookies for Adobe's login domain before reconnecting, or sign out completely from an Adobe tab in the same browser. This forces a profile selection prompt so you can choose the correct organization. |
Server not visible after adding | Verify the URL is exactly https://mcp.frame.io/v1/mcp — no trailing slash. Restarting your client after adding the server may be required. |
Tools not appearing in conversation | Verify the server shows as connected. If connected but tools are missing, check that the tool category is not set to Blocked in permissions. |
Connected but no workspaces or projects appear | Ask your AI agent to list all accounts first. If you have access to multiple, you may be in the wrong one. AMA users should verify they authenticated under the correct Adobe organization. |
Permission errors on specific tools | Tool calls reflect your Frame.io account permissions. Uploads require Contributor access or higher. |
Upload stalled or stuck in processing | Ask your AI agent: "What's the upload status of my recent file?" Large or high-resolution files can take several minutes to transcode. |
FAQ
Q: Which AI clients are supported?
A: Claude (Desktop, Web, iOS, Cowork, Code), ChatGPT (Desktop, Web, iOS), VS Code, and Cursor. See Anthropic's and OpenAI's connector documentation to confirm setup steps for your specific surface.
Q: Does the Frame.io MCP Server work with Legacy (V3) accounts?
A: No. This server requires a V4 account with an Adobe ID.
Q: Is my Frame.io data used to train AI models?
A: No. In line with Adobe's AI guidelines, customer data is never used to train or augment AI models. NLP is used only to parse the text query you enter — it has no awareness of your content. Semantic search generates embeddings stored securely in Adobe's internal database; only you can trigger a lookup against them.
Q: What actions can the AI agent take on my account?
A: The agent can only take actions your account permissions allow. Write actions (uploading, moving files, updating shares) require your explicit approval by default — keep the Needs approval setting until you've reviewed each tool type.
Q: Are there any tools that permanently delete assets without recovery?
A: No. Files moved to "recently deleted" can be restored.
Q: How do I disconnect my AI client?
A:
Claude Desktop: Go to Customize → Connectors, find Frame.io, and remove it.
ChatGPT: Go to Settings → Connected Apps and remove the Frame.io app.
VS Code: Remove the Frame.io entry from your MCP configuration.
Cursor: Remove the frameio-mcp entry from ~/.cursor/mcp.json.