Skip to main content
Glama
Cutover MCP

Cutover MCP

Official
by gocutover

Cutover MCP (Model Context Protocol) License: MIT

An MCP server for interacting with the Cutover API, powered by FastMCP.

Contents

Related MCP server: Maverick MCP Server

Capabilities

The server exposes 45 tools over MCP, grouped into 14 categories.

Tool Groups

Group

Description

runbooks

Create, read, update, and control the lifecycle of runbooks and templates

runbook_types

List, fetch, and query runbook types

tasks

Read, add, update, and progress (start/complete/skip/delete) tasks within a runbook

task_types

Look up task types, including integration-backed and agentic (AI agent) task types

streams

Create, read, update, and delete streams and substreams within a runbook

workspaces

List, search, fetch, and create workspaces

teams

Look up teams associated with a runbook

users

Look up and search for users

custom_fields

Discover custom fields and their metadata/valid options, and create, update and archive field definitions and groups

data_sources

Get a signed URL to upload the file a data source reads from

folders

List folders within a workspace

comments

Post and read comments on a runbook or task

activities

Read a runbook's activity trail

action_logs

Read the platform audit log

Tools

Runbooks

  • workspace_id: The unique identifier for the workspace (string, required)

  • is_template: Filter by template status — false excludes templates, true shows only templates (boolean, optional)

  • archived: Filter by archived status — false excludes archived runbooks, true shows only archived (boolean, optional)

  • source_runbook_id: Filter to only runbooks created from this template ID (string, optional)

  • folder_id: Filter to only runbooks in this folder ID (string, optional)

  • extra_params: Additional query parameters to pass to the API, e.g. {"stage": "active"} (object, optional)

  • runbook_id: The unique identifier for the runbook (string, required)

  • name: The name of the new runbook (string, required)

  • workspace_id: The workspace to create the runbook in — required unless copy_source_runbook_id is passed (string, optional)

  • description: Description for the runbook (string, optional)

  • status: RAG status — off, red, amber, green (string, optional)

  • is_template: Whether the runbook is a template (boolean, optional)

  • template_type: off, default, or snippet (string, optional)

  • rto: Recovery Time Objective in seconds (integer, optional)

  • timezone: IANA timezone name (string, optional)

  • runbook_type_id: Runbook type to associate (string, optional)

  • rto_start_task / rto_end_task: Start/end task IDs for the RTO/RTA feature (string, optional)

  • folder_id: Folder to place the runbook in — defaults to the workspace's default location (string, optional)

  • custom_field_values: Custom field values to set, including runbook dashboard fields — see Custom Fields (array of object, optional)

  • master_template: Whether this runbook can generate app-specific templates — requires is_template=true, and the API rejects it otherwise; on rejection, surface it rather than silently setting is_template=true (boolean, optional)

  • start_scheduled: ISO 8601 timestamp, or the literal "now", to schedule the start (string, optional)

  • end_scheduled: ISO 8601 timestamp for the scheduled end — must be omitted if start_scheduled isn't set (string, optional)

  • auto_start: Whether to auto-start once start_scheduled is reached (runs live, comms on) (boolean, optional)

  • copy_source_runbook_id: ID of an existing runbook/template to copy from (string, optional)

  • copy_tasks / copy_teams / copy_users: What to include in the copy — each defaults to true; teams copied without members unless copy_users is also true (boolean, optional)

  • shift_fixed_times: When copying, recalculate fixed task times relative to the new runbook's start (boolean, optional)

  • runbook_id: The unique identifier for the runbook (string, required)

  • name / description: New name/description (string, optional)

  • status: RAG status — off, red, amber, green (string, optional)

  • is_template: Whether the runbook is a template (boolean, optional)

  • rto: Recovery Time Objective in seconds (integer, optional)

  • timezone: IANA timezone name (string, optional)

  • rto_start_task / rto_end_task: Start/end task IDs for the RTO/RTA feature (string, optional)

  • custom_field_values: Custom field values to update, including runbook dashboard fields — see Custom Fields (array of object, optional)

  • folder_id: Folder to move the runbook to (string, optional)

  • master_template: Requires is_template=true; the API rejects it on a non-template runbook — surface the rejection instead of auto-flipping is_template (boolean, optional)

  • start_scheduled / end_scheduled: ISO 8601 timestamps (or "now" for start) to (re)schedule the runbook (string, optional)

  • auto_start: Whether to auto-start once start_scheduled is reached (boolean, optional)

  • runbook_id: The unique identifier for the runbook (string, required)

  • action: start, cancel, pause, or resume — the only supported actions (string, required)

  • comms: off, test, on — start only (string, optional, default "off")

  • disable_task_notify: Disable task start notifications — start only (boolean, optional)

  • run_type: live or rehearsal — start only (string, optional, default "rehearsal")

  • rebaseline: Recalculate all planned times based on current time — start only (boolean, optional)

  • shift_fixed_times: Shift fixed-time tasks relative to current time — start only (boolean, optional)

  • validation_level: warning or error — start only (string, optional, default "error")

  • message: Note attached to the action — cancel/pause/resume only (string, optional)

  • notify: Notify users about the action — cancel/pause/resume only (boolean, optional)

  • runbook_id: The template runbook ID to find copies of (string, required)

Runbook Types

No parameters.

  • runbook_type_id: The unique identifier for the runbook type (string, required)

  • query: Case-insensitive text matched against the runbook type name, key, and description (string, optional)

  • incident: Filter by whether the type is for incidents (boolean, optional)

  • enable_rto: Filter by whether the type supports the RTO/RTA feature (boolean, optional)

  • dynamic: Filter by whether the type is dynamic (boolean, optional)

  • ai_create_enabled: Filter by whether AI-assisted creation is enabled for the type (boolean, optional)

  • include_archived: Include archived runbook types; archived types are excluded by default (boolean, optional)

  • include_disabled: Include disabled runbook types; disabled types are excluded by default (boolean, optional)

Tasks

  • runbook_id: The unique identifier for the runbook (string, required)

  • forecast: Return all tasks with computed timing (start_display/end_display) and dependency graph fields, overriding pagination/filters (boolean, optional)

  • fields_task: Specific task fields to return, e.g. ["name", "stage"] (array of string, optional)

  • stage: Filter by stage — default, startable, in_progress, complete (array of string, optional)

  • stream_id: Filter to tasks in these stream IDs (array of string, optional)

  • completion_type: complete_normal, complete_skipped, complete_abandoned, complete_auto (string, optional)

  • task_type_id: Filter to tasks of these task type IDs (array of string, optional)

  • level: Filter by task level (string, optional)

  • search_term: Match against task name (string, optional)

  • has_comments: Only return tasks that have comments (boolean, optional)

  • runbook_team_id: Filter to tasks assigned to these runbook team IDs (array of string, optional)

  • user_id: Filter to tasks assigned to these user IDs (array of string, optional)

  • source_runbook_id: Filter to tasks originating from these source runbook IDs (array of string, optional)

  • sort: Sort field, e.g. start_planned or -start_planned for descending (string, optional)

  • runbook_id: The runbook to add the task to (string, required)

  • name: Task name (string, required)

  • description: Task description (string, optional)

  • task_type_id: Task type id from list_task_types, which includes integration-backed and agentic types (string, optional)

  • stream_id: Stream (or substream) to assign the task to (string, optional)

  • predecessors: Task IDs that are predecessors to this task (array of string, optional)

  • duration: Planned duration in seconds (integer, optional)

  • task_links: Links to other resources — link_type="runbook" links to a template runbook (target must be a template with ≥1 task and the task's task_type_id must be the tenant's linked type, or the link is silently dropped); link_type="snippet" attaches snippets (array of object, optional)

  • message: Message body for an Email/SMS/Call task, or initial prompt for an agentic task (string, optional)

  • recipients: Recipients for a comms task (array of object, optional)

  • assignees: Assignees to add — only users/teams already participating on the runbook are honored, others are silently ignored (array of object, optional)

  • custom_field_values: Custom field values to set — see Custom Fields (array of object, optional)

  • start_fixed / end_fixed: ISO 8601 timestamps fixing start/end (string, optional)

  • level: level_1, level_2, or level_3 (string, optional, default level_3)

  • auto_start: Start automatically once predecessors complete (boolean, optional)

  • auto_finish: Complete automatically once started (boolean, optional)

  • runbook_id / task_id: The runbook and task to update (string, required)

  • name / description: New name/description (string, optional)

  • predecessors: Task IDs that are predecessors to this task (array of string, optional)

  • task_type_id: Task type id from list_task_types, which includes integration-backed and agentic types (string, optional)

  • stream_id: Stream (or substream) to assign the task to (string, optional)

  • duration: Planned duration in seconds (integer, optional)

  • custom_field_values: Custom field values to update — see Custom Fields (array of object, optional)

  • assignees: Assignees to add — only existing runbook participants are honored; additive by default (array of object, optional)

  • delete_excluded_assignees: When true, replaces the full assignee list with only those provided instead of adding to it (boolean, optional, default false)

  • task_links: Replaces the task's links entirely — pass an empty list to clear all links (array of object, optional)

  • message: Message body for an Email/SMS/Call task, or initial prompt for an agentic task (string, optional)

  • recipients: Recipients for a comms task (array of object, optional)

  • start_fixed / end_fixed: ISO 8601 timestamps fixing start/end (string, optional)

  • level: level_1, level_2, or level_3 (string, optional)

  • auto_start / auto_finish: Auto-start on predecessor completion / auto-finish on start (boolean, optional)

  • runbook_id / task_id: The runbook and task to start (string, required)

  • runbook_id / task_id: The runbook and task to complete (string, required)

  • runbook_id / task_id: The runbook and task to skip (string, required)

  • comment: Reason for skipping — required by the API and posted as a runbook comment (string, required)

  • runbook_id / task_id: The runbook and task to delete (string, required)

Task Types

  • include_integration: Include task types backed by an integration action (boolean, optional, default true)

  • integration_action_id: The integration action id from the task type's integration_action relationship (string, required)

Streams

  • runbook_id: The runbook to list streams for (string, required)

  • forecast: Include computed forecast fields (start_display, end_display, etc.) (boolean, optional)

  • runbook_id / stream_id: The runbook and stream to fetch (string, required)

  • runbook_id: The runbook to create the stream in (string, required)

  • name: Stream name (string, required)

  • description: Stream description (string, optional)

  • color: CSS-friendly color, e.g. #f0f0f0 or rgb(0,0,0) (string, optional)

  • parent_stream_id: Parent stream ID, if creating a substream (string, optional)

  • runbook_id / stream_id: The runbook and stream to update (string, required)

  • name / description / color: New values (string, optional)

  • runbook_id / stream_id: The runbook and stream to delete (string, required)

Workspaces

  • limit: Maximum workspaces to return (integer, optional, default 50)

  • offset: Number of workspaces to skip (integer, optional, default 0)

  • query: Search string (string, required)

  • workspace_id: The unique identifier for the workspace (string, required)

  • name: Workspace/account name (string, required)

  • key: Shortened version of the name (string, required)

  • description: Workspace/account description (string, optional)

Teams

  • runbook_id: The runbook to fetch teams for (string, required)

Users

  • user_id: The user ID to fetch (string, required)

  • query: Search string — name or email, partial match (string, required)

Custom Fields

The runbook and task tools accept custom_field_values as a list of {"custom_field_id": "123", "value": ...} entries, where the id comes from list_custom_fields. A field's internal name can be used in place of the id, but its display_name cannot. Each value is a string, or a list of strings for multi-select fields. Malformed entries are rejected before any request is sent to Cutover.

  • workspace_id: Filter to a workspace — omit to return fields from all accessible workspaces (string, optional)

  • include_global: When workspace_id is set, also include globally available fields alongside the workspace's own (boolean, optional, default true)

  • scope: task, runbook, or all (string, optional, default "all") — runbook also covers dashboard fields on the runbook homepage, PIR and Incident Review

  • custom_field_id: The custom field to retrieve (string, required)

The tools below change field definitions (the Custom Fields admin page), not the values on a runbook or task. They need the Custom Fields Admin or Global Admin role.

  • name: The field's name (string, required). The API sets the UI label to the same value

  • field_type: text, textarea, select_menu, radiobox, checkboxes, datetime or user_select (string, required)

  • apply_to: runbook_add_edit, runbook_edit, runbook_page, task_add_edit, task_edit, task_start or task_end (string, required)

  • workspace_id: The workspace that owns the field (string, required unless global_field is true)

  • global_field: Create the field for every workspace (boolean, optional, default false)

  • field_options: Option names — required for select_menu, radiobox and checkboxes, rejected otherwise (array of string, optional)

  • required, allow_field_creation, display_list, display_search, display_table: Field settings (boolean, optional)

  • default_value: The value a new runbook or task starts with (string, optional)

  • custom_field_group_id: The group to show the field in (string, optional)

  • task_type_ids, runbook_type_ids: Limit the field to these task or runbook types (array of string, optional)

  • custom_field_id: The field to update (string, required)

  • name: The new name (string, optional)

  • add_field_options: Option names to add to a select_menu, radiobox or checkboxes field; existing options are kept, and other field types are rejected (array of string, optional)

  • required, allow_field_creation, display_list, display_search, display_table: Field settings (boolean, optional)

  • default_value: The value a new runbook or task starts with (string, optional)

  • custom_field_group_id: Move the field to this group, or "" to take it out of its group; omitted, the field stays in its current group (string, optional)

  • task_type_ids, runbook_type_ids: Replace the constraints of that type; the other type is kept, and an empty list is rejected because the API cannot clear them (array of string, optional)

  • custom_field_id: The field to archive (string, required)

Takes no parameters. Archived groups are not returned; the API does not list them.

  • name: The group's name (string, required)

  • custom_field_group_id: The group to rename (string, required)

  • name: The new name (string, required)

  • custom_field_group_id: The group to archive (string, required)

Data Sources

  • file_name: Name of the file to upload, e.g. cmdb.csv (string, required)

  • data_source_id: An existing data source to upload a new version of the file for; omit for the first upload (integer, optional)

The server cannot read files on your machine, so the agent uploads the file itself with an HTTP PUT to the returned url before it expires (expires_in seconds), then creates the data source with the returned datasource_endpoint_url as its URL. A new version for an existing data source is only read if the returned datasource_endpoint_url is that data source's current URL; refresh and sync the data source afterwards. Fails with a clear error if the instance's data source search service is not configured.

Folders

  • workspace_id: The workspace to list folders for (string, required)

Comments

  • runbook_id: The runbook to comment on (string, required)

  • content: Comment text — a limited set of HTML tags is supported (e.g. <p>, <b>, <ul>, <code>); markdown is not rendered and disallowed tags are stripped (string, required)

  • task_id: Task to attach the comment to — omit to post at runbook level (string, optional)

  • runbook_id: The runbook to read comments from (string, required)

  • task_id: Return only the comments attached to this task — omit for every comment on the runbook, including runbook-level ones (string, optional)

Activities

  • runbook_id: The runbook to retrieve activities for (string, required)

  • created_after / created_before: ISO 8601 date bounds (string, optional)

Action Logs

  • runbook_id / user_id / workspace_id: Filter by runbook, user, or workspace (string, optional)

  • created_after / created_before: ISO 8601 date bounds (string, optional)

  • max_pages: Maximum pages to fetch — response flags truncated: true if more remain (integer, optional, default 10)

This includes both read and write tools — see Security & Responsible Use for how to scope access appropriately.

Installation

The server is published on PyPI as cutover-mcp and runs with uv, which downloads Python 3.13+ on demand, so uv is the only prerequisite. On macOS:

brew install uv

Other platforms: see the uv installation guide.

Then point your MCP client at uvx cutover-mcp with your credentials in its env block — see Client Setup. uvx downloads the latest release on first run and caches it; pin a version with uvx cutover-mcp@1.2.3. No clone, no virtualenv.

To start it from a terminal instead (it waits for an MCP client on stdin, so it will look idle; Ctrl+C to stop):

CUTOVER_BASE_URL=https://api.cutover.net CUTOVER_API_TOKEN=your_token_here CUTOVER_CORE_URL=https://YOUR-INSTANCE-NAME.cutover.com uvx cutover-mcp

Setup

Only needed to work on the server itself or to run an unreleased build; to just use it, see Installation above.

  1. Copy .env.example to .env and fill in your Cutover API credentials — see Environment Configuration below.

  2. Install dependencies with uv:

    uv sync
  3. Run the server using:

    uv run cutover-mcp

    That is the same console script uvx cutover-mcp runs; uv run python src/cutover_mcp/server.py works too.

  4. Run tests:

    uv run pytest

Environment Configuration

Which URLs do I need?

For most instances, one URL and a token:

CUTOVER_CORE_URL=https://your-instance.cutover.com
CUTOVER_API_TOKEN=MY_TOKEN_VALUE

CUTOVER_CORE_URL is your Cutover instance URL, the one you open in the browser. The API host (CUTOVER_BASE_URL) defaults to https://api.cutover.net, which serves every shared production instance, so leave it unset unless your instance has its own API host (single-tenant deployments, typically https://api.your-instance.cutover.com). The default only applies to Cutover-hosted instance URLs; for any other instance URL the server asks you to set CUTOVER_BASE_URL rather than guess. If a call fails with an authorization error the message says how to override it; ask your Cutover contact if you are unsure which applies.

Environment Variables Reference

Variable

Required

Notes

CUTOVER_CORE_URL

Yes

Your Cutover instance URL, e.g. https://your-instance.cutover.com. Sent as the Core-Url header so the API routes to your instance.

CUTOVER_API_TOKEN

Yes

See Generating an API Token below.

CUTOVER_BASE_URL

No

The Cutover API host. Defaults to https://api.cutover.net. Set it only for an instance with its own API host, e.g. https://api.your-instance.cutover.com.

Generating an API Token

There are two ways to generate a token:

  • Access Management — an admin generates a token for a specific user (useful if you want the token to act as someone other than yourself, e.g. a global admin, or you don't have self-service access).

  • My Details → User App Tokens — generate your own token directly, self-service.

Security & Responsible Use

The MCP server adds no security model of its own — it authenticates to the Cutover API as a single user app token over HTTPS and inherits exactly that user's permissions. Scope the token correctly and use the write tools deliberately:

  • Least privilege. Use a dedicated non-interactive user with the Developer role plus only the scoped roles the use case needs (e.g. Workspace manager on one workspace) — not a personal admin token. Rotate the token periodically and revoke it in Cutover when the use case ends.

  • Read-only vs read-write. This server ships write tools (create_runbook, manage_runbook, delete_task, start_task, …) and does not enforce read-only itself. For query-only use cases, omit roles that grant create/update/delete/manage — least privilege is configured in Cutover, not the client.

  • Approve writes. Keep human approval enabled for mutating or consequential actions — models can act on unexpected content returned by tools, so approval is the real safeguard.

  • Verify critical output. A tool call returns real data, but the model's summary of it can still be wrong (or answer with no tool call at all). For compliance- or audit-critical output, verify against Cutover as the system of record.

  • Protect the token. It lives in env config (.env* is git-ignored and excluded from Docker images). If a token is compromised, revoke it immediately in Cutover's Access Management — the MCP server holds no independent session, so revoking the token fully cuts off access, and a scoped role confines any impact to that use case.

Project Structure

  • src/cutover_mcp/ - Main package

  • clients/ - API client

  • resources/ - Resource definitions (MCP endpoints)

  • tools/ - Tool logic

  • server.py - FastMCP server entrypoint (main() is the cutover-mcp console script)

  • server.json - The server's MCP Registry entry

  • tests/ - Tests

Notes

  • Dependency and environment management is handled by uv and hatchling.

  • Versioned releases and the changelog live under Releases. Each release is published to PyPI and listed in the MCP Registry as io.github.gocutover/cutover-mcp. The server reports its version as serverInfo.version and from GET /health when running over HTTP.

Client Setup

Run in VS Code GitHub Copilot

If you want to use VS Code with GitHub Copilot to call out to this MCP server, you can use mcp.json.example.

  1. Copy mcp.json.example to .vscode/mcp.json (create the .vscode/ directory if not already present). It runs uvx cutover-mcp and prompts for your API token the first time the server starts, so the token is never written into the project.

  2. Set CUTOVER_BASE_URL and CUTOVER_CORE_URL in its env block for your instance — see Environment Configuration.

  3. Restart VS Code for the changes to take effect. In the Copilot Chat box, click the Configure tools icon and expand cutover-mcp to see its full list of tools. MCP tools only run in Agent mode — switch the mode dropdown from Ask to Agent if nothing happens.

Run in Claude Desktop

If you'd rather use Claude Desktop instead of an IDE, you can set that up:

  1. Download and install Claude Desktop.

  2. Copy claude_desktop_config.json.example to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS path shown; on Windows, use %APPDATA%\Claude\claude_desktop_config.json).

  3. Fill in the env block with your credentials — see Environment Configuration. Claude Desktop does not inherit your shell's PATH, so if it reports that uvx cannot be found, set command to the full path from which uvx (typically /opt/homebrew/bin/uvx on macOS).

  4. Restart Claude Desktop. Click the + button at the bottom of the chat box, then Connectors — cutover-mcp should be listed there along with its tools. If it isn't, check Settings → Developer for its connection status and logs.

This will allow Claude Desktop to connect to your MCP server.

Docker Usage

Docker is optional — the server runs fine without it (see Installation). Use this if you want a self-contained way to run it instead of uvx.

From the repository root, build the image:

docker build -t cutover-mcp .

.env is excluded from the build via .dockerignore, so the image isn't tied to any environment or credential — never bake .env into an image you build, push, or share.

Pass your .env file in at run time instead, with --env-file:

docker run -i --rm --env-file .env cutover-mcp

Example: Using Docker in MCP Server Configuration

You can configure your MCP server to use the Docker container instead of uvx. The root key differs by client: VS Code (mcp.json) uses servers; Claude Desktop (claude_desktop_config.json) uses mcpServers.

VS Code (.vscode/mcp.json):

{
  "servers": {
    "cutover-mcp": {
      "command": "docker",
      "args": [
        "run",
        "-i",
        "--rm",
        "--env-file",
        "/path/to/your/repo/.env",
        "cutover-mcp"
      ]
    }
  }
}

Claude Desktop (claude_desktop_config.json):

{
  "mcpServers": {
    "cutover-mcp": {
      "command": "docker",
      "args": [
        "run",
        "-i",
        "--rm",
        "--env-file",
        "/path/to/your/repo/.env",
        "cutover-mcp"
      ]
    }
  }
}

To use Docker in the example configuration files (mcp.json.example and claude_desktop_config.json.example), replace the existing commands with the Docker-based commands shown in the "Docker Usage" section above.

FAQs

How can I start the MCP server via Visual Studio Code?

Open up the .vscode/mcp.json file. Hover above the name of the server i.e. 'cutover-mcp'. You should have the option to start the server.

Troubleshooting

Why is VS Code not giving me the ability to start the MCP server?

Ensure you aren't running the server independently in another terminal. Close and re-open VS Code, and if that doesn't work, restart your machine.

Alternatively, you can use the following approach to start the server:

View -> Command Palette -> MCP: List Servers -> your mcp server -> start/restart/stop server.

Available Tools

45 tools
add_commentAdd CommentA

Post a comment on a runbook, optionally attached to a specific task.

Use this to record progress notes or results (e.g. while working through a runbook's tasks) without overwriting task descriptions.

Inputs: content (required), runbook_id (required), task_id. Returns: data, included

add_task_to_runbookAdd Task To RunbookC

Add a new task to an existing runbook.

Inputs: name (required), runbook_id (required), level, message, +13 more. Returns: data, included

archive_custom_fieldArchive Custom FieldA

Archive a custom field so it can no longer be used. Existing values stay on runbooks and tasks. Requires the Custom Fields Admin role (or Global Admin).

Inputs: custom_field_id (required). Returns: object

archive_custom_field_groupArchive Custom Field GroupA

Archive a custom field group. Requires the Custom Fields Admin role (or Global Admin).

Inputs: custom_field_group_id (required). Returns: object

complete_taskComplete TaskC

Complete a specific task in a runbook.

Inputs: task_id (required), runbook_id (required). Returns: data, included

create_custom_fieldCreate Custom FieldA

Create a custom field definition (the Custom Fields admin page). This defines a new field; to set a value on a runbook or task, use custom_field_values on the runbook and task tools.

Requires the Custom Fields Admin role (or Global Admin). Check list_custom_fields first so an existing field is reused rather than duplicated.

Inputs: name (required), apply_to (required), field_type (required), required, +11 more. Returns: object

create_custom_field_groupCreate Custom Field GroupA

Create a custom field group. Groups are shared across the instance. Check list_custom_field_groups first so an existing group is reused. Requires the Custom Fields Admin role (or Global Admin).

Inputs: name (required). Returns: object

create_runbookCreate RunbookA

Create a new runbook in a workspace, or as a copy of an existing runbook/template.

Inputs: name (required), rto, status, timezone, +18 more. Returns: data, included

create_streamCreate StreamC

Create a new stream or substream in a runbook.

Inputs: name (required), runbook_id (required), color, description, +1 more. Returns: data, included

create_workspaceCreate WorkspaceC

Create a new account/workspace.

Inputs: key (required), name (required), description. Returns: object

delete_streamDelete StreamC

Delete a stream from a runbook.

Inputs: stream_id (required), runbook_id (required). Returns: object

delete_taskDelete TaskC

Delete a single task from a runbook.

Inputs: task_id (required), runbook_id (required). Returns: object

get_action_logsGet Action LogsB

Retrieve action logs (audit logs) from Cutover. Paginates through results up to max_pages.

Inputs: user_id, max_pages, runbook_id, workspace_id, +2 more. Returns: object

get_activitiesGet ActivitiesB

Get activities for an incident runbook to track recent actions and events.

Inputs: runbook_id (required), created_after, created_before. Returns: result

get_commentsGet CommentsA

List the comments on a runbook, optionally filtered to a specific task.

Comments are the append-only, timestamped log channel while a runbook is running (task descriptions are locked outside edit/dynamic mode), so this is how to read back progress notes or status that were recorded earlier with add_comment, by a person or by an agent.

Inputs: runbook_id (required), task_id. Returns: data, meta, links, included

get_custom_fieldGet Custom FieldB

Get a custom field's metadata including its type and valid options.

Inputs: custom_field_id (required). Returns: object

get_data_source_upload_urlGet Data Source Upload UrlA

Get a short-lived, pre-signed URL for uploading a file that a data source reads from, such as a CMDB export in JSON or CSV.

This server cannot read files on your machine, so upload the file yourself, straight away: send an HTTP PUT with the raw file as the body to the returned url before it expires (expires_in seconds). For example:

curl --fail -X PUT --upload-file ./cmdb.csv "<url>"

Keep the url in double quotes, because it contains & characters the shell would otherwise split on. Keep --fail, so a rejected upload exits with an error instead of looking like a success. If the upload fails or the url has expired, call this tool again for a new url rather than reusing the old one.

For a new data source, create it with datasource_endpoint_url as its url. The file is scanned for malware after upload, so test the data source can read it before importing.

Inputs: file_name (required), data_source_id. Returns: object

get_integration_task_type_detailsGet Integration Task Type DetailsA

Get how an integration-backed task type is configured, to explain or choose it.

Call this only when list_task_types does not make a type's purpose clear, typically a custom/payload type or one of several similar types. Pass the id from the task type's integration_action relationship. The result includes the action's name, kind, title, descriptions, hooks and its descriptive settings (target URLs, request type, response mode, polling, custom field mappings); credentials are never included.

Reading integration actions needs the global Read IntegrationAction capability, held by global admins, integrations admins, global stakeholders and workspace stakeholders. Access to the runbook does not grant it, so runbook admins and workspace managers get a 403. On a 403, go by the task type's name and kind instead and tell the user why.

Inputs: integration_action_id (required). Returns: object

get_runbook_by_idGet Runbook By IdB

Fetch details for a specific runbook by its ID.

Inputs: runbook_id (required). Returns: data, included

get_runbook_tasksGet Runbook TasksA

Fetch tasks for a specific runbook, with optional filtering and sorting.

This endpoint has two modes:

  • Conventional mode: supports filtering, pagination, and sorting.

  • Forecast mode (forecast=true): returns ALL tasks in a single response regardless of pagination or filters, and includes computed timing fields start_display and end_display, plus predecessor/successor graph relationships. Use for timeline or dependency views.

Inputs: runbook_id (required), sort, level, stage, +10 more. Returns: data, meta, links, included

get_runbook_teamsGet Runbook TeamsB

Get all teams associated with a runbook.

Inputs: runbook_id (required). Returns: result

get_runbook_template_copiesGet Runbook Template CopiesA

Get all runbooks that were created from a specific runbook template.

Inputs: runbook_id (required). Returns: data, meta, links, included

get_runbook_type_by_idGet Runbook Type By IdC

Fetch details for a specific runbook type by its ID.

Inputs: runbook_type_id (required). Returns: data, included

get_streamGet StreamB

Get details of a specific stream or substream in a runbook.

Inputs: stream_id (required), runbook_id (required). Returns: data, included

get_userGet UserC

Get user details by user ID.

Inputs: user_id (required). Returns: object

get_workspace_by_idGet Workspace By IdB

Fetch details for a specific workspace by its ID.

Inputs: workspace_id (required). Returns: object

list_custom_field_groupsList Custom Field GroupsA

List custom field groups, which control the section a custom field is shown in. Archived groups are not returned; the API does not list them.

Inputs: no named parameters. Returns: result

list_custom_fieldsList Custom FieldsB

List all available custom fields to discover fields that may not have values yet.

Inputs: scope, workspace_id, include_global. Returns: result

list_foldersList FoldersC

List all folders in a specific workspace.

Inputs: workspace_id (required). Returns: data, meta, links, included

list_runbooksList RunbooksB

List all runbooks in a specific workspace.

Inputs: workspace_id (required), archived, folder_id, is_template, +2 more. Returns: data, meta, links, included

list_runbook_typesList Runbook TypesA

List all runbook types in the instance.

Inputs: no named parameters. Returns: data, meta, links, included

list_streamsList StreamsB

List all streams for a specific runbook, including substreams.

Inputs: runbook_id (required), forecast. Returns: data, meta, links, included

list_task_typesList Task TypesA

List all task types, including the ones backed by an integration action.

Use this to find the task_type_id for a task type by name ("Normal", "Milestone", "Email", ...) and to find the right type when a task should do something: run a Jenkins build, raise a ServiceNow change, post to Slack, call a webhook, or hand the task to an AI agent (an "agentic", "AI", "Cutover AI" or chat task).

A task type backed by an integration carries a meta.integration_action object: kind (<provider>/<action>, e.g. jenkins/build_job, service_now/change_request, custom/payload, agentic/chat_cutover), a definition with the title and description of that kind of action as declared by the integration, hooks (when it fires, usually task/started) and a description entered by the admin (often missing); plus integration_action / integration relationships, and its name is the integration action's name. Work out what a type does from the kind, definition, description and name. custom/payload types are admin-built HTTP integrations whose purpose is only in their name; call get_integration_task_type_details to see how one is configured when it matters.

Agentic types: prefer agentic/chat_cutover (the built-in Cutover AI agent) over agentic/chat (a customer-hosted agent) unless the user names one. For an agentic task set the task's message to the initial prompt the agent should receive when the task starts.

Inputs: include_integration. Returns: object

list_workspacesList WorkspacesC

List all workspaces.

Inputs: limit, offset. Returns: object

manage_runbookManage RunbookB

Manage a specific runbook by performing an action (start, cancel, pause, resume). These are the only possible actions with this tool.

Inputs: action (required), runbook_id (required), comms, notify, +6 more. Returns: object

query_runbook_typesQuery Runbook TypesA

Find runbook types matching a name/description search and/or capability flags.

Use this to resolve a runbook type a user described in words into the id that create_runbook needs — e.g. "an incident runbook" (incident=True) or "a template with RTO/RTA enabled" (enable_rto=True). Every criterion is optional; passing none returns all usable runbook types. Multiple criteria are combined with AND.

Filtering is applied client-side over the full collection because the public API ignores query parameters on core/runbook_types.

Inputs: query, dynamic, incident, enable_rto, +3 more. Returns: data, meta, links, included

query_workspacesQuery WorkspacesC

Search for workspaces by name or query string.

Inputs: query (required). Returns: object

search_usersSearch UsersB

Search for users by name or email (fuzzy match).

Inputs: query (required). Returns: result

skip_taskSkip TaskC

Skip a specific task in a runbook.

Inputs: comment (required), task_id (required), runbook_id (required). Returns: object

start_taskStart TaskC

Start a specific task in a runbook.

Inputs: task_id (required), runbook_id (required). Returns: data, included

update_custom_fieldUpdate Custom FieldA

Update a custom field definition. Only the parameters you pass are changed. The field's type, apply_to and workspace cannot be changed; create a new field instead.

Requires the Custom Fields Admin role (or Global Admin).

Inputs: custom_field_id (required), name, required, display_list, +8 more. Returns: object

update_custom_field_groupUpdate Custom Field GroupB

Rename a custom field group. Requires the Custom Fields Admin role (or Global Admin).

Inputs: name (required), custom_field_group_id (required). Returns: object

update_runbookUpdate RunbookC

Update a specific runbook's fields.

Inputs: runbook_id (required), rto, name, status, +11 more. Returns: data, included

update_runbook_taskUpdate Runbook TaskC

Update an existing task in a runbook (including dependencies, description, stream, duration, etc.).

Inputs: task_id (required), runbook_id (required), name, level, +15 more. Returns: data, included

update_streamUpdate StreamC

Update an existing stream in a runbook.

Inputs: stream_id (required), runbook_id (required), name, color, +1 more. Returns: data, included

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    B3.1/5.0

    Scored across 45 tools

    Disambiguation4/5

    Most tools target distinct resource/action pairs, and the descriptions clarify boundaries between similar operations. There is only mild potential confusion between get_activities and get_action_logs, and between list/query variants for workspaces and runbook types.

    Naming Consistency4/5

    Tool names are consistently snake_case and mostly follow a verb_noun pattern. Minor deviations include get_*_by_id versus get_stream/get_user, manage_runbook, and add_task_to_runbook, but the naming remains readable and predictable.

    Tool Count2/5

    With 45 tools, the surface is heavy for a single server and exceeds the range where tools are easily scannable. Although the Cutover domain is broad, the many CRUD clusters for custom fields, groups, streams, and types increase selection burden significantly.

    Completeness3/5

    Core runbook, task, stream, comment, and custom-field workflows are well covered. However, there are notable gaps: no delete/archive runbook operation, no workspace update/delete, folders are list-only, users are read-only, and data-source tooling only provides an upload URL.

    Maintenance

    ActivityActive
    ResponsivenessLimited data

    Related MCP Connectors

    Related MCP Servers

    • -
      license
      Not graded
      quality
      Not graded
      maintenance
      An MCP server that enables Amazon Q CLI users to create, query, and manage Maverick sites through natural language commands.
      -
    • A
      license
      Not graded
      quality
      B
      maintenance
      MCP server for interacting with Keepit backup and data protection services. Enables monitoring, management, and security operations through natural language.
      5
      MIT
    • A
      license
      C
      quality
      B
      maintenance
      An MCP server that provides complete Kubernetes API operations, enabling cluster management, resource CRUD, diagnostics, backup/restore, and multi-cluster configuration via natural language.
      32
      2
      MIT