REST API (legacy v3 endpoints)

The Server AI Toolkit provides REST API endpoints for AI agent tools.

Authentication

All requests must include:

• Authorization: Bearer <JWT>
• X-App-Id: <APP_ID>

Generate the JWT server-side with your AI secret. Get your App ID and secret key on the Server AI Toolkit settings page.

import jwt from 'jsonwebtoken'

const JWT_TOKEN = jwt.sign(
  {
    experimental_document_server_id: 'your-document-server-id',
    experimental_document_server_management_api_secret:
      'your-document-server-management-api-secret',
  },
  'your-ai-secret-key',
  { expiresIn: '1h' },
)

See the authorization guide for more information.

Document Server credentials

Include these JWT claims when you want the Server AI Toolkit to fetch and save Tiptap Cloud documents automatically:

• experimental_document_server_id
• experimental_document_server_management_api_secret

Base URL

The hosted base URL is:

https://api.tiptap.dev

In the examples below, requests are written as BASE_URL/v3/ai/....

How to provide the document

Tool endpoints accept exactly one document source:

• document: Inline Tiptap JSON
• experimental_documentOptions: A Tiptap Cloud document reference

experimental_documentOptions has the following shape:

• documentId (string, required): The Tiptap Cloud document identifier.
• userId? (string | null, optional): Identifier attributed to edits the AI performs.
• field? (string | null, optional, default: "default"): Targets a specific collaborative field (Y.js XML fragment) inside the document when omitted or null. Use this when your document stores multiple editable fields — for example a separate title and body — under the same documentId.

Example with a Tiptap Cloud document:

{
  "experimental_documentOptions": {
    "documentId": "your-document-id",
    "userId": "ai-assistant"
  }
}

Example targeting a non-default collaborative field:

{
  "experimental_documentOptions": {
    "documentId": "your-document-id",
    "userId": "ai-assistant",
    "field": "title"
  }
}

Example with an inline document:

{
  "document": {
    "type": "doc",
    "content": []
  }
}

Formats

Most endpoints accept a format field:

• "json": Standard JSON representation
• "shorthand": Tiptap Shorthand, a token-efficient format for prompts and model output

Tool endpoints

Get tool definitions

POST /v3/ai/toolkit/tools

Returns the available tool definitions for AI agents.

Request body:

• editorContext (EditorContext, required). Editor context obtained from the getEditorContext function. See installation guide.
• tools? (Record<string, boolean>, optional): Enable or disable individual tools. See the tools reference for the supported keys.
• operationMeta? (string, optional, default: "")
• format? ('json' | 'shorthand', optional, default: 'json')
• schemaAwarenessData? (SchemaAwarenessData, optional, deprecated): Use editorContext instead. Supported for backward compatibility.

Response:

• prompt (string): Add this to the system prompt for your AI request. It teaches the AI how the Tiptap document works, what elements it can contain, and how the document format works.
• tools (array)
  • name (string)
  • description (string)
  • inputSchema (object): JSON schema for the AI-generated tool input.

Execute a tool

POST /v3/ai/toolkit/execute-tool

Executes one Server AI Toolkit tool such as tiptapRead, tiptapEdit, getThreads, or editThreads. See the tools reference for every supported tool and its type shapes.

Request body:

• toolName (string, required)
• input (required): Tool input generated by the AI model. The shape depends on toolName; see the tools reference.
• editorContext (EditorContext, required unless schemaAwarenessData is provided)
• schemaAwarenessData? (SchemaAwarenessData, optional, deprecated): Use editorContext instead. Supported for backward compatibility.
• toolConfig? (optional): Tool-specific parameters provided by the developer, not by the AI model. See the tools reference.
• format? ('json' | 'shorthand', optional, default: 'json')
• reviewOptions? (ReviewOptions, optional, default: { mode: 'disabled' })
• experimental_commentsOptions? ({ threadData?: Record<string, unknown>, commentData?: Record<string, unknown> }, optional): Metadata attached to new threads and comments created during thread-related tool execution.
• exactly one of:
  • document
  • experimental_documentOptions

Response:

• output: tool response that should be read by the AI model. See the tools reference.
• toolResult: tool response in a format that is designed to be parsed by the developer. Should not be read by the AI. See the tools reference.
• docChanged (boolean): whether the document changed as a result of the tool call
• document (object | null): the document after the AI made changes to it

Stream a tool

POST /v3/ai/toolkit/stream-tool

Streams tool execution. The request shape matches POST /v3/ai/toolkit/execute-tool. The streaming response format is still being finalized.

Because this endpoint works on the document over a live connection, it can return connection and concurrency errors that the REST endpoints don't. They arrive as an error event in the response stream — see Streaming and session errors.

Legacy workflow endpoints

Get a workflow definition

Use one of these endpoints:

• POST /v3/ai/toolkit/workflows/edit
• POST /v3/ai/toolkit/workflows/insert-content
• POST /v3/ai/toolkit/workflows/proofreader
• POST /v3/ai/toolkit/workflows/threads

Request body:

• format? ('json' | 'shorthand', optional, default: 'json'). The proofreader workflow only accepts 'shorthand'.

Response:

• systemPrompt (string)
• outputSchema (object)

Example:

curl --location 'BASE_URL/v3/ai/toolkit/workflows/edit' \
  --header 'Content-Type: application/json' \
  --header 'Authorization: Bearer YOUR_JWT_TOKEN' \
  --header 'X-App-Id: YOUR_APP_ID' \
  --data '{
    "format": "shorthand"
  }'

Read document content for a workflow

POST /v3/ai/toolkit/read/read-document

Reads document content and returns workflow-ready content.

Request body:

• range? ({ from: number, to: number }, optional, default: entire document)
• schemaAwarenessData (SchemaAwarenessData, required)
• format? ('json' | 'shorthand', optional, default: 'json')
• chunkSize? (number, optional, default: 32000)
• reviewOptions? (ReviewOptions, optional, default: { mode: 'disabled' })
• exactly one of:
  • document
  • experimental_documentOptions

Response:

• output
  • success (boolean)
  • content? (unknown)
  • error? (string)
• docChanged (boolean)
• document (object | null)

Example:

curl --location 'BASE_URL/v3/ai/toolkit/read/read-document' \
  --header 'Content-Type: application/json' \
  --header 'Authorization: Bearer YOUR_JWT_TOKEN' \
  --header 'X-App-Id: YOUR_APP_ID' \
  --data '{
    "schemaAwarenessData": { /* schema awareness data */ },
    "format": "shorthand",
    "reviewOptions": {
      "mode": "disabled"
    },
    "experimental_documentOptions": {
      "documentId": "your-document-id",
      "userId": "ai-assistant"
    }
  }'

Read the current selection

POST /v3/ai/toolkit/read/read-selection

Reads the selection and returns either a selection payload or an explicit empty state.

Request body:

• range ({ from: number, to: number }, required)
• schemaAwarenessData (SchemaAwarenessData, required)
• format? ('json' | 'shorthand', optional, default: 'json')
• chunkSize? (number, optional, default: 32000)
• reviewOptions? (ReviewOptions, optional, default: { mode: 'disabled' })
• exactly one of:
  • document
  • experimental_documentOptions

Response:

• output
  • when the selection is empty:
    • isEmpty (true)
  • when the selection contains content:
    • isEmpty (false)
    • content (unknown)
    • nodeHashes (string[])
    • nodeRange ({ from: number, to: number })
    • prompt (string)
• docChanged (boolean)
• document (object | null)

Example:

curl --location 'BASE_URL/v3/ai/toolkit/read/read-selection' \
  --header 'Content-Type: application/json' \
  --header 'Authorization: Bearer YOUR_JWT_TOKEN' \
  --header 'X-App-Id: YOUR_APP_ID' \
  --data '{
    "schemaAwarenessData": { /* schema awareness data */ },
    "range": { "from": 10, "to": 42 },
    "format": "shorthand",
    "experimental_documentOptions": {
      "documentId": "your-document-id"
    }
  }'

Read threads

POST /v3/ai/toolkit/read/threads

Reads all threads and comments for a Tiptap Cloud document.

Request body:

• schemaAwarenessData (SchemaAwarenessData, required)
• format ('json' | 'shorthand', required)
• experimental_documentOptions ({ documentId: string, userId?: string, field?: string | null }, required)

Response:

• output
  • threads? (unknown[])
  • error? (string)
• docChanged (boolean)
• document (object | null)

Example:

curl --location 'BASE_URL/v3/ai/toolkit/read/threads' \
  --header 'Content-Type: application/json' \
  --header 'Authorization: Bearer YOUR_JWT_TOKEN' \
  --header 'X-App-Id: YOUR_APP_ID' \
  --data '{
    "schemaAwarenessData": { /* schema awareness data */ },
    "format": "shorthand",
    "experimental_documentOptions": {
      "documentId": "your-document-id",
      "userId": "ai-assistant"
    }
  }'

Execute the insert-content workflow

POST /v3/ai/toolkit/execute-workflow/insert-content

Request body:

• input (unknown, required): Generated insert-content payload
• range? ({ from: number, to: number }, optional)
• schemaAwarenessData (SchemaAwarenessData, required)
• format? ('json' | 'shorthand', optional, default: 'json')
• chunkSize? (number, optional, default: 32000)
• reviewOptions? (ReviewOptions, optional, default: { mode: 'disabled' })
• exactly one of:
  • document
  • experimental_documentOptions

Response:

• output
  • error? (string)
  • range? ({ from: number, to: number })
• docChanged (boolean)
• document (object | null)

Example:

curl --location 'BASE_URL/v3/ai/toolkit/execute-workflow/insert-content' \
  --header 'Content-Type: application/json' \
  --header 'Authorization: Bearer YOUR_JWT_TOKEN' \
  --header 'X-App-Id: YOUR_APP_ID' \
  --data '{
    "schemaAwarenessData": { /* schema awareness data */ },
    "input": "This is the replacement content.",
    "range": { "from": 10, "to": 42 },
    "format": "shorthand",
    "reviewOptions": {
      "mode": "disabled"
    },
    "experimental_documentOptions": {
      "documentId": "your-document-id",
      "userId": "ai-assistant"
    }
  }'

Execute the Tiptap Edit workflow

POST /v3/ai/toolkit/execute-workflow/tiptap-edit

Request body:

• input (object, required): Edit workflow operations
• schemaAwarenessData (SchemaAwarenessData, required)
• format? ('json' | 'shorthand', optional, default: 'json')
• chunkSize? (number, optional, default: 32000)
• reviewOptions? (ReviewOptions, optional, default: { mode: 'disabled' })
• exactly one of:
  • document
  • experimental_documentOptions

Response:

• output
  • operationResults? (array)
  • reason? ('validationError' | 'unexpectedError')
  • error? (string)
• docChanged (boolean)
• document (object | null)

Example:

curl --location 'BASE_URL/v3/ai/toolkit/execute-workflow/tiptap-edit' \
  --header 'Content-Type: application/json' \
  --header 'Authorization: Bearer YOUR_JWT_TOKEN' \
  --header 'X-App-Id: YOUR_APP_ID' \
  --data '{
    "schemaAwarenessData": { /* schema awareness data */ },
    "input": {
      "operations": [
        {
          "type": "replace",
          "target": "abc123",
          "content": "# Updated heading\n\nUpdated paragraph content."
        }
      ]
    },
    "format": "shorthand",
    "reviewOptions": {
      "mode": "disabled"
    },
    "experimental_documentOptions": {
      "documentId": "your-document-id",
      "userId": "ai-assistant"
    }
  }'

Execute the proofreader workflow

POST /v3/ai/toolkit/execute-workflow/proofreader

Request body:

• input (object, required): Proofreader operations
• schemaAwarenessData (SchemaAwarenessData, required)
• format ('shorthand', required)
• chunkSize? (number, optional, default: 32000)
• reviewOptions? (ReviewOptions, optional, default: { mode: 'disabled' })
• exactly one of:
  • document
  • experimental_documentOptions

Response:

• output
  • operationResults (array)
• docChanged (boolean)
• document (object | null)

Example:

curl --location 'BASE_URL/v3/ai/toolkit/execute-workflow/proofreader' \
  --header 'Content-Type: application/json' \
  --header 'Authorization: Bearer YOUR_JWT_TOKEN' \
  --header 'X-App-Id: YOUR_APP_ID' \
  --data '{
    "schemaAwarenessData": { /* schema awareness data */ },
    "input": {
      "operations": []
    },
    "format": "shorthand",
    "reviewOptions": {
      "mode": "trackedChanges",
      "trackedChangesOptions": {
        "userId": "ai-assistant"
      }
    },
    "experimental_documentOptions": {
      "documentId": "your-document-id"
    }
  }'

Execute the comments workflow

POST /v3/ai/toolkit/execute-workflow/threads

Request body:

• input (object, required): Thread operations
• schemaAwarenessData (SchemaAwarenessData, required)
• format? ('json' | 'shorthand', optional, default: 'json')
• experimental_documentOptions ({ documentId: string, userId?: string, field?: string | null }, required)
• experimental_commentsOptions? ({ threadData?: Record<string, unknown>, commentData?: Record<string, unknown> }, optional): Metadata attached to new threads and comments created by the workflow.

Response:

• output
  • operations? (array)
  • error? (string)
• docChanged (boolean)
• document (object | null)

Example:

curl --location 'BASE_URL/v3/ai/toolkit/execute-workflow/threads' \
  --header 'Content-Type: application/json' \
  --header 'Authorization: Bearer YOUR_JWT_TOKEN' \
  --header 'X-App-Id: YOUR_APP_ID' \
  --data '{
    "schemaAwarenessData": { /* schema awareness data */ },
    "input": {
      "operations": [
        {
          "type": "createComment",
          "threadId": "thread-123",
          "content": "Please clarify this point."
        }
      ]
    },
    "format": "shorthand",
    "experimental_documentOptions": {
      "documentId": "your-document-id",
      "userId": "ai-assistant"
    },
    "experimental_commentsOptions": {
      "threadData": {
        "source": "ai"
      },
      "commentData": {
        "source": "ai"
      }
    }
  }'

Error handling

REST endpoints return errors with this shape:

{
  "error": {
    "message": "Human-readable description of what went wrong.",
    "status": 502,
    "code": "document_load_failed"
  }
}

• message (string): Human-readable description. Don't match on it — the wording can change.
• status (number): The HTTP status code, also set on the response.
• code (string): A stable identifier you can branch on. Validation failures use code: "validation_failed" and add an issues array.

The API uses standard HTTP error codes:

• 400 Bad Request: Invalid body, invalid format, missing document source, or missing Document Server credentials
• 401 Unauthorized: Missing or invalid JWT or App ID
• 404 Not Found: Unknown endpoint or tool
• 422 Unprocessable Entity: Validation failed
• 429 Too Many Requests: Duplicate request or rate-limit protection
• 500 Internal Server Error: Unexpected server error
• 502 Bad Gateway: Failed to load or save the Tiptap Cloud document

Streaming and session errors

POST /v3/ai/toolkit/stream-tool works on a Tiptap Cloud document over a live connection: it joins the document in real time so the AI edits the latest content alongside other collaborators. Errors that happen while establishing or using that connection are delivered in-band as an error event in the response stream, not as an HTTP status:

{ "type": "error", "code": "concurrent_edit_conflict", "status": 409, "message": "..." }

The status field mirrors the equivalent HTTP status. Possible code values:

Handle concurrent edits

concurrent_edit_conflict is expected in collaborative documents. While the AI reads a document, calls the model, and writes its changes, a user can edit the same document in between. Rather than overwrite that edit, the toolkit rejects the write so no work is lost.

Treat it as a safe, retriable signal: when you receive an error event with code: "concurrent_edit_conflict", run the same tool call again. The toolkit re-reads the document, so the AI works from the user's latest changes. Cap the number of retries to avoid long loops on a document that's being edited heavily.

// `runStream` sends the start/delta/end messages and resolves to the stream's
// final `error` event, or `null` when the stream completes successfully.
async function streamToolWithRetry(body: unknown, maxRetries = 3) {
  for (let attempt = 0; attempt <= maxRetries; attempt++) {
    const error = await runStream(body)

    if (!error) {
      return // completed successfully
    }
    if (error.code !== 'concurrent_edit_conflict') {
      throw new Error(`stream-tool failed: ${error.code}`)
    }
    // Document changed mid-flight — retry with the latest content.
  }

  throw new Error('Tool call kept conflicting with concurrent edits.')
}

Available tools

Below is a list of the available tools.

tiptapRead

Reads a slice of top-level document nodes. It also edits the document to prepare it for efficient edits.

Tool config (toolConfig)

null

Tool input (toolInput)

• from (number): Zero-based index of the first top-level node to read.

Tool output (toolOutput)

Success response:

• success (true): Indicates that the read request was successful.
• totalNodeCount (number): Total number of top-level nodes in the document.
• nodeRange ([number, number]): Top-level node range that was read.
• content (JSONContent[] | string): Document content for the returned range. When format is json, this is a Tiptap JSON fragment. When format is shorthand, this is a Tiptap Shorthand string.

Error response:

• success (false): Indicates that the read request failed.
• error (string): Error message.
• totalNodeCount? (number): Total number of top-level nodes.

Tool result (toolResult)

null

tiptapEdit

Edits document nodes by replacing a target, inserting before a target, or inserting after a target. Targets are node hashes from tiptapRead, or "doc" for the whole document.

Tool config (toolConfig)

null

Tool input (toolInput)

• operations (array): Edit operations applied sequentially. The format of the operations is described in the tool definition. If you are a customer, contact us for more information on the operation format.

Tool output (toolOutput)

Success or partial-success response:

• success (boolean): Whether all operations completed successfully.
• operationResults (array): Result for each operation.
  • success (boolean): Whether the operation completed successfully.
  • target (string): String identifier of the element targeted by the operation.
  • error (string | null): Failure reason, or null if the operation succeeded.

Error response:

• success (false): Indicates that the edit request failed.
• reason ('validationError' | 'unexpectedError'): Error category.
• error (string): Error message.

Tool result (toolResult)

null

getThreads

Reads persisted comment threads, their comments, and their document locations.

Tool config (toolConfig)

null

Tool input (toolInput)

• from (number): Zero-based index of the first thread to read.

Tool output (toolOutput)

Success response:

• success (true): Indicates that the read request was successful.
• totalThreadCount (number): Total number of threads in the document.
• threadRange ([number, number]): Thread range that was read, as [from, to).
• threads (array): Returned thread objects.
  • id (string): Unique thread identifier.
  • nodeRange ([number, number] | null): Node range where the thread is located, or null if the thread is not annotated in the document.
  • content (JSONContent[] | string | null): Content marked by the thread. When format is json, this is a Tiptap JSON fragment. When format is shorthand, this is a Tiptap Shorthand string. null means the thread is not annotated in the document.
  • comments (array): Comments in the thread.
    • id (string): Unique comment identifier.
    • content (string): Comment text.
    • userId (string): ID of the user who created the comment.
    • createdAt (string): ISO timestamp when the comment was created.
    • updatedAt (string): ISO timestamp when the comment was last updated.
  • resolvedAt? (string | null): ISO timestamp when the thread was resolved, or null if unresolved.
  • createdAt (string): ISO timestamp when the thread was created.
  • updatedAt (string): ISO timestamp when the thread was last updated.
  • data? (Record<string, unknown>): Public thread metadata.

Error response:

• success (false): Indicates that the read request failed.
• error (string): Error message.
• totalThreadCount? (number): Total number of persisted threads, when known.

Tool result (toolResult)

null

editThreads

Creates, updates, removes, resolves, or unresolves comment threads and comments.

Tool config (toolConfig)

null

To attach metadata to new AI-generated threads or comments, pass experimental_commentsOptions on the execute-tool request.

Tool input (toolInput)

• operations (array): Thread operations applied sequentially. The format of the operations is described in the tool definition. If you are a customer, contact us for more information on the operation format.

Tool output (toolOutput)

Success or partial-success response:

• success (boolean): Indicates whether all operations completed successfully.
• operations (array): Result for each operation.
  • type (string): Operation type that was executed.
  • success (boolean): Whether the operation completed successfully.
  • message (string): Human-readable operation result.
  • threadId? (string): Thread ID for operations that create or reference a thread.

Error response:

• success (false): Indicates that the request failed before operations were processed.
• error (string): Error message.

Tool result (toolResult)

null

readDocument

Reads the content of the entire document.

Tool config (toolConfig)

null

Tool input (toolInput)

{}

Tool output (toolOutput)

Success response:

• success (true): Indicates that the document was read successfully.
• content (JSONContent[] | string): Whole effective document content. When format is json, this is a Tiptap JSON fragment. When format is shorthand, this is a Tiptap Shorthand string.

Tool result (toolResult)

Same shape as toolOutput.

proofreader

Applies small edits to the document. Optimized for use cases where the AI makes small edits of one or few words. For example, proofreading the document or checking for spelling mistakes.

This tool only supports format: "shorthand".

Tool config (toolConfig)

null

Tool input (toolInput)

• operations (array): Proofreader operations applied sequentially. The format of the operations is described in the tool definition. If you are a customer, contact us for more information on the operation format.

Tool output (toolOutput)

Success or partial-success response:

• success (boolean): Indicates whether all operations completed successfully.
• operationResults (array): Result for each operation.
  • target (string): String identifier of the element targeted by the operation.
  • success (boolean): Whether the operation completed successfully.
  • error (string | null): Failure reason, or null when the operation succeeded.

Tool result (toolResult)

null

Related pages

• Review options
• Editor context