Adobe Design Data Agent MCP, Adobe's token authoring server

• Loads the full token and component overview

  Returns the token taxonomy, resolved values, component list and field definitions. The server says to call it at the start of a session.

  Try: “Load the design data primer.”

  Read · primer

• Finds tokens with a filter expression

  Returns the design tokens that match a filter expression.

  Try: “Find all tokens for the button component.”

  Read · query_tokens

• Gives a token's final value in context

  Resolves a token property to its final value for a component, variant, state or theme setting, and flags ambiguous or deprecated results.

  Try: “What is the background color of the primary button in dark mode?”

  Read · resolve_token

• Suggests tokens from a plain description

  Returns tokens matching a description in plain words, ranked by confidence.

  Try: “Which token fits a subtle divider line?”

  Read · suggest_token

• Shows a component's schema and token bindings

  Returns the JSON schema and token relationships for a component by its ID.

  Try: “Describe the checkbox component and its tokens.”

  Read · describe_component

• Reads one design guideline

  Returns the full content and metadata of a guideline by its ID.

  Try: “Show the motion guideline.”

  Read · describe_guideline

• Lists the design guidelines

  Lists guideline IDs and metadata, optionally by category.

  Try: “List the guidelines about color.”

  Read · list_guidelines

• Checks token usage in a dataset

  Runs structural and relational checks on a dataset and returns a report of violations and warnings.

  Try: “Validate the tokens in my dataset and list the errors.”

  Files · validate_usage

• Compares two datasets

  Returns the tokens that were added, removed or changed between two datasets.

  Try: “Compare my old and new token folders and list what changed.”

  Files · diff_datasets

• Saves design context into the dataset

  Writes agent-generated design context, such as a product context file, into the dataset.

  Try: “Save a summary of our brand rules into the dataset.”

  Write · write

• Starts a guided session to add a new token

  Opens a token authoring session that you step through, and returns a session ID.

  Try: “Start adding a new border color token.”

  Write · start_authoring_session

• Describes the token you want and gets matches

  Records what the token is for and returns existing tokens that could be reused.

  Try: “The token is for the focus ring color. Is there one already?”

  Write · authoring_session_step_intent

• Sets the token's layer, property and name

  Sets the layer, property and name fields for the new token and checks them against the field catalog.

  Try: “Classify it as a component token for background-color.”

  Write · authoring_session_step_classification

• Sets the token's values

  Sets value rows that each hold a literal value or an alias, for the modes that matter.

  Try: “Set the light value to #fff and dark to #000.”

  Write · authoring_session_step_values

• Writes the new token to a file

  Builds the token, writes it into a tokens file and ends the session.

  Try: “Commit the token to the color tokens file.”

  Write · authoring_session_commit

• Cancels a token authoring session

  Cancels a session and deletes its saved file.

  Try: “Cancel the current authoring session.”

  Delete · authoring_session_cancel

• Shows a session's current state

  Returns the current state of an authoring session.

  Try: “Where am I in the current authoring session?”

  Read · authoring_session_get

• Lists open authoring sessions

  Lists every active authoring session.

  Try: “List my open authoring sessions.”

  Read · authoring_session_list

• Changes fields on an existing token

  Merges updates into a committed token and checks it again. The token's ID is kept.

  Try: “Change the description of the focus ring token.”

  Write · edit_token

• Marks a token as deprecated

  Adds a deprecation marker, and can record a replacement and a planned removal version.

  Try: “Deprecate the old accent token and point to the new one.”

  Write · deprecate_token

• Gives a token a new name

  Assigns a new name to a token and keeps its ID, with an optional pointer to the replacement.

  Try: “Rename the border token to border-subtle.”

  Write · rename_token

• Points an alias token at another token

  Changes which token an alias refers to. The new target must exist.

  Try: “Make the link color alias point to the new accent token.”

  Write · rewire_alias

• Deletes a token from a file

  Deletes a token, but stops if any other token still refers to it.

  Try: “Remove the unused legacy-gray token.”

  Delete · remove_token

• Adds a mode to a mode set

  Appends a new mode to an existing mode-set file.

  Try: “Add a high-contrast mode to the color scheme set.”

  Write · add_mode

• Renames a mode everywhere it is used

  Renames a mode in its mode-set file and updates every token file that uses it.

  Try: “Rename the dark mode to night.”

  Write · rename_mode

• Deletes a mode from a mode set

  Removes a mode, but stops if any token uses it or if it is the default.

  Try: “Remove the unused sepia mode.”

  Delete · remove_mode

• Creates a new mode set

  Writes a new mode-set file for a new dimension, such as a density option.

  Try: “Create a density mode set with compact and comfortable.”

  Write · create_mode_set

• Deletes a mode set

  Deletes a mode-set file, but stops if any token still uses that dimension.

  Try: “Remove the old platform mode set.”

  Delete · remove_mode_set

• Creates a component, field, guideline or similar file

  Creates a new non-token data object after checking it against its schema. Fails if the file exists.

  Try: “Create a new guideline page for data tables.”

  Write · data_create

• Replaces an existing data file

  Replaces a non-token data object with a new document after checking it against its schema. It is a full replace, not a patch.

  Try: “Update the typography guideline with this text.”

  Write · data_edit

Watch out for

• It edits and deletes files on disk

  Tools can remove tokens, modes and mode sets and replace whole data files. Commit before you start so you can review and undo. github.com

• Removals have built-in guards

  Removing a token, mode or mode set stops if something else still refers to it. Check what depends on an item before you ask for a rename. github.com

• Point it at the right folder

  Path settings decide which dataset it edits. Set them to a project checkout, not a folder you cannot restore. github.com

Security advisories