Sanity MCP Server: Setup for Claude, Cursor and Other Apps

• Runs a GROQ query on your content

  Runs a GROQ query against a dataset and returns the matching documents. Results over 64 KiB are rejected, so ask for only the fields you need.

  Try: “Show all articles written by Mark, titles only.”

  Read · query_documents

• Fetches one document by its ID

  Looks up a single document by its exact ID and returns all of it. It does not search.

  Try: “Open the document with ID article-1.”

  Read · get_document

• Shows the deployed content schema

  Fetches the schema deployed for a project and dataset, so the assistant knows your document types before it queries or edits.

  Try: “What document types does my production dataset have?”

  Read · get_schema

• Lists every deployed schema, by source

  Shows each deployed schema for a dataset, grouped by where it came from, with an ID to read an exact one.

  Try: “Which schemas are deployed to this dataset?”

  Read · list_workspace_schemas

• Reads one type from the schema

  Returns one document or object type from the current schema in editable form, with the version number later edits must pass.

  Try: “Show me the definition of the article type.”

  Read · schema_type_get

• Checks documents against the schema

  Validates documents against the deployed schema without changing them. It is in beta, so review the results.

  Try: “Validate my draft articles against the schema.”

  Read · documents_validate

• Shows which Sanity account is connected

  Returns the signed-in account name, email and how the server is authenticated, to confirm the right identity.

  Try: “Which Sanity account am I connected as?”

  Read · whoami

• Lists your Sanity organizations

  Returns the organizations your account can access.

  Try: “Which Sanity organizations do I belong to?”

  Read · list_organizations

• Lists your Sanity projects

  Returns the projects linked to your account.

  Try: “List my Sanity projects.”

  Read · list_projects

• Lists Studios linked to a project

  Returns the Studio applications attached to one project.

  Try: “Which Studios does this project have?”

  Read · get_project_studios

• Lists a project's datasets

  Returns the datasets in one project.

  Try: “What datasets are in this project?”

  Read · list_datasets

• Lists content releases in a dataset

  Shows active and scheduled releases by default, and published or archived ones when you filter by state.

  Try: “Show all releases in production.”

  Read · list_releases

• Lists allowed web origins for a project

  Returns the CORS origins that may make browser requests to the project.

  Try: “Which CORS origins are allowed on this project?”

  Read · cors_origins_list

• Lists search indexes for a dataset

  Returns the embeddings indexes available for meaning-based search.

  Try: “Which semantic search indexes exist here?”

  Read · list_embeddings_indices

• Searches content by meaning

  Runs a semantic search on an embeddings index and returns the closest documents.

  Try: “Find articles about remote work policies.”

  Read · semantic_search

• Searches Sanity's documentation

  Looks up pages in the Sanity docs by keyword.

  Try: “Search the Sanity docs for GROQ joins.”

  Network · search_docs

• Reads one Sanity docs article

  Fetches the full text of a named documentation article.

  Try: “Read the docs page on releases.”

  Network · read_docs

• Lists Sanity's coding best-practice rules

  Lists the development rule sets the server can load for your assistant.

  Try: “Which Sanity best-practice rules are available?”

  Read · list_sanity_rules

• Loads chosen best-practice rules

  Returns the text of specific rule sets, for example for writing GROQ, so the assistant follows them.

  Try: “Load the GROQ rules before you write the query.”

  Read · get_sanity_rules

• Creates new documents

  Creates documents from the content you give. They become drafts, live documents for types with live edit, or release versions when a release is named.

  Try: “Create a draft article titled Spring launch.”

  Write · create_documents

• Edits existing documents

  Applies edits to up to 25 documents per call in single transactions. Changes go to drafts or release versions, never straight to published content.

  Try: “Change the title of article-1 to New title.”

  Write · patch_documents

• Adds a document to a release

  Creates a release-specific copy of a document, used before you patch content that belongs to a release.

  Try: “Add article-1 to the autumn release.”

  Write · create_version

• Publishes drafts

  Publishes the drafts you name, which makes the content live. Documents are checked against the schema first.

  Try: “Publish the draft article-1.”

  Write · publish_documents

• Takes published documents back to drafts

  Unpublishes documents so they stop being live and return to draft state.

  Try: “Unpublish the old pricing page.”

  Write · unpublish_documents

• Deletes draft documents

  Throws away the drafts you name and keeps the published versions. The draft edits are lost.

  Try: “Discard my draft of article-1.”

  Delete · discard_drafts

• Removes documents from a release

  Discards document versions from a release, which deletes those release copies.

  Try: “Drop article-1 from the autumn release.”

  Delete · version_discard

• Creates a content release

  Makes a new release for grouping changes. It does not schedule or publish it; that happens in Studio or the HTTP API.

  Try: “Create a release called Autumn campaign.”

  Write · create_release

• Deploys schema types to the cloud

  Directly deploys schema types to the project, creating or replacing the schema the server manages.

  Try: “Deploy an article type with title and body fields.”

  Write · deploy_schema

• Adds or edits fields on a type

  Adds, replaces or removes top-level fields on one document or object type, using its current version number.

  Try: “Add a slug field to the article type.”

  Write · schema_fields_patch

• Removes types from the schema

  Deletes one or more types from the dataset schema. Types still used by other types cannot be removed, and a dry run shows losses first.

  Try: “Remove the unused promo type from the schema.”

  Delete · schema_types_delete

• Removes a server-managed schema

  Removes a legacy schema the server manages in a named workspace. Documents and hosted Studios are kept.

  Try: “Undeploy the old schema in the marketing workspace.”

  Delete · undeploy_schema

• Deploys a hosted Sanity Studio

  Creates a hosted Studio bound to the server-managed schema and returns its address. Run it again after schema changes.

  Try: “Deploy a Studio for this schema.”

  Write · deploy_studio

• Creates a new Sanity project

  Makes a project with a dataset and API tokens.

  Try: “Create a Sanity project called Docs site.”

  Write · create_project

• Creates a new dataset

  Adds a dataset with the name and access setting you choose.

  Try: “Create a private dataset called staging.”

  Write · create_dataset

• Changes a dataset's visibility

  Switches a dataset between public and private and sets its description. Public datasets can be read without sign-in.

  Try: “Make the staging dataset private.”

  Write · update_dataset

• Allows a web origin

  Adds origins that may make browser requests to the project.

  Try: “Allow https://example.com to call this project.”

  Write · add_cors_origin

• Removes an allowed web origin

  Deletes a CORS origin, so browsers from that site lose access.

  Try: “Remove the old staging origin.”

  Delete · cors_origins_delete

• Generates an image for a document field

  Starts AI image generation for a field. It uses your Sanity AI credits.

  Try: “Generate a hero image for article-1.”

  Write · generate_image

• Transforms an existing image with AI

  Starts an AI change to an image already in your content. It uses AI credits.

  Try: “Make the cover image of article-1 look like a watercolor.”

  Write · transform_image

• Uploads a file from a web link to a dataset

  Starts an upload of an image or file from a public HTTPS link and returns an operation ID to track it.

  Try: “Upload this logo URL to my dataset.”

  Write · dataset_assets_upload_from_url

• Prepares a one-use upload link for a dataset

  Returns a temporary address where the client posts file bytes of up to 50 MiB into a dataset.

  Try: “Upload the attached PDF to my dataset.”

  Write · dataset_assets_upload_from_file

• Uploads a ChatGPT file to a dataset

  Fetches a ChatGPT web attachment or generated file from its temporary link and stores it in a dataset.

  Try: “Add the image I just generated to my dataset.”

  Write · dataset_assets_upload_from_web_chatgpt

• Checks progress of asset uploads

  Looks up one to 20 uploads by operation ID and returns asset IDs, links and references for up to 24 hours.

  Try: “Is my logo upload done?”

  Read · assets_upload_status

• Lists Media Libraries in an organization

  Returns library IDs to pass to the other Media Library tools.

  Try: “Which Media Libraries does my organization have?”

  Read · media_library_list

• Shows a Media Library's types

  Returns the document types of a library, or the fields of one type, before you query or edit.

  Try: “What fields do assets have in this library?”

  Read · media_library_get_schema

• Queries a Media Library with GROQ

  Searches assets, collections, folders and aspect definitions in a library.

  Try: “Find all PNG assets added this month.”

  Read · media_library_query_documents

• Fetches one Media Library document

  Looks up a single library document by its exact ID.

  Try: “Open asset image-abc in the library.”

  Read · media_library_get_document

• Shows where an asset is used

  Lists documents across project datasets that reference a library asset, so you can check before replacing it.

  Try: “Where is this logo used?”

  Read · media_library_list_asset_references

• Creates collections in a Media Library

  Creates asset collections. Assets themselves cannot be created this way.

  Try: “Create a collection called Press photos.”

  Write · media_library_create_documents

• Edits Media Library metadata

  Changes asset titles, aspects and CDN access policy, and collection details. Title and aspect edits stay as drafts until published.

  Try: “Retitle this asset to Team photo 2026.”

  Write · media_library_patch_documents

• Publishes Media Library asset edits

  Publishes draft edits to assets after checking them against the library's aspect definitions.

  Try: “Publish my changes to the Team photo asset.”

  Write · media_library_publish_documents

• Discards Media Library asset drafts

  Throws away unpublished metadata edits and keeps the published asset.

  Try: “Discard my draft edits on this asset.”

  Delete · media_library_discard_drafts

• Makes library assets usable in documents

  Returns field values that reference chosen library assets, ready to apply with the document edit tool. Image and file assets get a local copy.

  Try: “Link this library image into article-1.”

  Write · media_library_link_assets

• Uploads a file from a web link to a library

  Starts an upload of an image, video or file from a public HTTPS link into a Media Library, or adds a version to an asset.

  Try: “Upload this video link to the Media Library.”

  Write · media_library_upload_from_url

• Prepares a one-use upload link for a library

  Returns a temporary address where the client posts file bytes of up to 50 MiB into a Media Library.

  Try: “Upload the attached photo to the Media Library.”

  Write · media_library_upload_from_file

• Old name for the library upload tool

  Marked deprecated by Sanity. It does not start an upload, and points you to the library upload tool.

  Try: “Upload this link to the Media Library.”

  Read · media_assets_upload_from_url

• Runs a limited set of Sanity CLI commands

  Runs approved Sanity commands without a shell and returns their output. It cannot touch your files or sign-in.

  Try: “Run the Sanity help command to list options.”

  Runs code · run_sanity_cli

• Sends feedback about Sanity to Sanity

  Submits a message to Sanity about an error, a missing feature or unclear docs.

  Try: “Tell Sanity that the schema tool error was confusing.”

  Write · give_sanity_feedback

Watch out for

• It can delete drafts, release items and schema types

  Discard, version-discard, schema type delete, undeploy and CORS delete tools remove real data or settings. Approve each call. sanity.io

• It can publish and change live content

  The publish tool makes content live, and the project tools create projects and datasets and switch datasets to public. Review each call before you accept it. sanity.io

• No read-only mode is documented

  Sanity lists no setting to turn off writes. If you use an API token, a token with a viewer role limits what the tools can do. sanity.io

• Two tools use AI credits

  The image generation and image transform tools use Sanity AI credits. You can turn them off in your client. sanity.io