MCP tools

The APIANT MCP server at https://mcp.apiant.ai/mcp gives Claude the tools below. You do not call these tools yourself: you tell Claude what you want in plain English, and Claude picks and calls the tools.

This list is generated from the APIANT.ai source code on 2026-09-15. It leaves out tools reserved for APIANT staff and for administrators who manage other accounts. A listed tool can still refuse a call when your plan or permissions do not allow it; the notes under each family say which checks apply. The In the assistant column marks the tools the in-app assistant also uses.

Family	What it covers
Automations (auto_)	Build, edit, commit, turn on and off, organize and version automations.
Runs and execution history (exec_)	Test automations, read run history and per-step data, and retry, replay, restart, halt or cancel runs.
Connections (connection_)	Connect and disconnect app accounts, with an API key or through an OAuth sign-in.
App catalog (catalog_)	Search apps, triggers, actions and fields, and read dropdown choices and dynamic fields while an automation is being built.
Assemblies: connectors, actions and triggers (asm_)	Create and edit the connectors, actions, triggers, dropdowns and field lookups that automations use, and test them against a live connection.
Error alerts (alert_)	Read and change the rules that rewrite error-alert messages for a step or an automation, read the system-wide rules, and manage dismissed alerts.
Publish (publish_)	Copy an automation, a folder of automations, or a connector, action or trigger from your dev environment to its prod environment.
Credentials for your own app registration (credential_)	Report how Claude can obtain credentials for an app, and return the callback URL and scopes to register on the vendor's developer portal.
Other tools	Check the connection to the MCP server, report the installed plugin version, and file a bug report or feedback with APIANT.

Automations (auto_)

Build, edit, commit, turn on and off, organize and version automations. Edits happen on a draft that is committed when it is ready.

Tool	What it does	In the assistant
`auto_apply_edits`	Apply several edits to an automation draft in order, stopping at the first one that fails.	Yes
`auto_build`	Create a complete automation in one call from a name, its trigger and an ordered list of steps.	Yes
`auto_commit`	Save the draft to the automation so it becomes the version that runs, optionally with a new name, folder, description or polling schedule.	Yes
`auto_compare_versions`	Show what changed between two saved versions of an automation, step by step.	No
`auto_discard_draft`	Throw away the current draft and leave the saved automation unchanged.	Yes
`auto_edit_mapping`	Change how data is mapped into a step's fields in the draft.	Yes
`auto_edit_metadata`	Rename a step or change its display name and other descriptive details in the draft.	Yes
`auto_edit_structure`	Add, remove, move or replace a step in the draft, including conditions, loops and subroutine calls.	Yes
`auto_edit_trigger`	Add, remove or replace a trigger in the draft, or change its filter.	Yes
`auto_find_by_assembly`	Find your automations that use a given action or trigger.	No
`auto_folder`	List, create, rename and delete automation folders, and move automations between them.	Yes
`auto_get_connection_throttle`	Read the rate limit set on one of your app connections.	No
`auto_get_draft_state`	Return the current draft of the automation being edited.	Yes
`auto_get_step_details`	Return one step of the draft by its position.	Yes
`auto_get_subroutine_io`	Return the inputs a subroutine expects and the outputs it returns.	No
`auto_get_subroutine_references`	Find the automations in your account, and in its linked child accounts, that call a subroutine or use it as an error handler.	No
`auto_get_summary`	Summarize an automation: its name, whether it is on, how it is triggered, its schedule and its step types.	Yes
`auto_get_version`	Read one saved version of an automation.	No
`auto_list`	List your automations, most recently edited first, filtered by name, folder, or on and off.	Yes
`auto_list_versions`	List the saved versions of an automation, newest first.	No
`auto_load`	Open a saved automation as a draft for editing.	Yes
`auto_manage`	Finish an automation: rename it, turn it on or off, set how many of its runs may overlap, detach a deployed copy from its template, or delete it.	Yes
`auto_pending_change_requests`	Collect the step comments you left on automation diagrams in the web app, so Claude can apply them.	Yes
`auto_set_active`	Turn an automation on or off. Turning it on is refused while a required setting is empty or an app it uses is not connected.	Yes
`auto_set_connection_throttle`	Set or clear a rate limit on one of your app connections, shared by every action that uses it.	No

Runs and execution history (exec_)

Test automations, read run history and per-step data, and retry, replay, restart, halt or cancel runs.

Lists that span automations (failed runs, waiting runs, pending retries, unprocessed webhooks, history search) cover your own automations.

Tool	What it does	In the assistant
`exec_cancel_snoozed`	Cancel an automation's snoozed or waiting runs so they never resume.	No
`exec_get_branch_test_points`	List an automation's condition branches and which of them a test run covered.	Yes
`exec_get_data`	Show the input, output and error of each step for one record processed in a run, including how each field value was mapped.	Yes
`exec_get_form_link`	Return the form link for a run that is waiting for someone to submit a hosted form.	Yes
`exec_get_history`	List an automation's runs, newest first, filtered by outcome, recent days or error text.	Yes
`exec_get_retry_detail`	Show one pending retry: its attempt number, its schedule and the failed step's error.	Yes
`exec_get_step_data`	Show each step of a run with its status, output, error and duration.	Yes
`exec_halt`	Stop one run of an automation, or every run of it that is in progress.	No
`exec_list_failed_runs`	List failed runs across your automations for a date range, newest first.	Yes
`exec_list_parked`	List runs that are waiting (snoozed, awaiting a form or an approval, or waiting on other runs) and what will resume each one.	No
`exec_list_retries`	List failed steps that are waiting for an automatic retry, soonest first.	No
`exec_list_unprocessed_webhooks`	List automations that received webhooks that were never processed, for example because the automation was off or a monthly limit was reached.	No
`exec_mutate_data`	Change the recorded output of a step in a run, so the run can be restarted with corrected data.	No
`exec_replay_webhook`	Run an automation again as a new run, with a JSON payload you supply or with an earlier run's trigger data.	No
`exec_restart_from_step`	Run a run again from a chosen step, reusing the recorded output of the steps before it.	No
`exec_retry_failed`	Run failed runs again as new runs, using each one's original trigger data.	Yes
`exec_search_history`	Search the run history of your automations for text in automation names, errors, step data or outgoing requests.	No
`exec_test_automation`	Run an automation as a test and return the result when it finishes.	Yes
`exec_test_subroutine`	Run a subroutine as a test with inputs you supply and return its outputs.	No

Connections (connection_)

Connect and disconnect app accounts, with an API key or through an OAuth sign-in.

The assistant does not use the two OAuth tools. It connects OAuth apps through a full-page redirect in the browser instead.

Tool	What it does	In the assistant
`connection_connect_api_key`	Connect an app with an API key or the other credentials the app asks for.	Yes
`connection_disconnect`	Disconnect a connected app account, removing webhooks it registered with the app for your automations.	Yes
`connection_finalize_oauth`	Finish connecting an app through its OAuth sign-in and save the connection.	No
`connection_start_oauth`	Start connecting an app through its OAuth sign-in and return the app's authorization link.	No

App catalog (catalog_)

Search apps, triggers, actions and fields, and read dropdown choices and dynamic fields while an automation is being built.

Downloading an app logo needs the Assembly Developer permission.
Repairing a shared action or trigger needs a plan that allows app authoring.

Tool	What it does	In the assistant
`catalog_classify_operation`	Check whether an app and operation you asked for are ready to use or need to be built or tried first, so Claude knows how to proceed.	No
`catalog_download_app_logo`	Fetch an app's logo by its web domain and set it as the connector's icon.	No
`catalog_get_app_operations`	List the actions or triggers of one app.	Yes
`catalog_get_dropdown_options`	Return the choices for a dropdown field, read from the app with your connection.	Yes
`catalog_get_dynamic_fields`	Return the fields an action or trigger offers for your connected account, such as custom fields.	Yes
`catalog_get_operation_fields`	Return the input and output fields of an action or trigger.	Yes
`catalog_list_apps`	List the apps in the catalog, including connectors your account built, and whether you have connected each one.	Yes
`catalog_repair_operation`	Give your account a private copy of a shared action or trigger that is broken, move your automation onto the copy so it can be fixed, and send the diagnosis to APIANT.	No
`catalog_reveal_operation`	Return the details of a newer action or trigger that the app picker does not show yet, so an automation can use it with your connection.	No
`catalog_search_apps`	Search the app catalog.	Yes
`catalog_search_fields`	Search the input and output fields of actions and triggers by name, label or help text.	Yes
`catalog_search_operations`	Search actions and triggers across the catalog.	Yes

Assemblies: connectors, actions and triggers (asm_)

Create and edit the connectors, actions, triggers, dropdowns and field lookups that automations use, and test them against a live connection.

Creating an action, trigger or connector, and finishing or repairing one, need a plan that allows app authoring.
Creating an API-key or no-auth connector, and setting an app icon, also need the Assembly Developer permission.
Creating an OAuth connector needs the OAuth App Registration permission or a plan that allows app registration.
You can change only the actions, triggers and connectors you built.

Tool	What it does	In the assistant
`asm_compare_versions`	Show which parts of an action, trigger or connector changed between two saved versions.	No
`asm_create_action_input_fields`	Set the input fields of an action: names, data types, required flags and help text.	No
`asm_create_action_or_trigger`	Create an action or trigger on a connector from an HTTP request and a mapping of the response.	No
`asm_create_connector`	Create a connector for an app that uses an API key or no credentials.	No
`asm_create_field_discovery`	Create a lookup that reads an app's field list from its API, and attach it to an action or trigger.	No
`asm_create_oauth1a_connector`	Create a connector for an app that signs in with OAuth 1.0a.	No
`asm_create_oauth_connector`	Create a connector for an app that signs in with OAuth 2.0.	No
`asm_create_populated_dropdown`	Create a dropdown whose choices come from the app's API, and attach it to an action or trigger.	No
`asm_delete_connector`	Delete a connector you built, once nothing else uses it.	No
`asm_delete_operation`	Delete an action or trigger you built.	No
`asm_document_fields`	Edit the label, help text and required flag of existing fields on an action or trigger, one field at a time.	No
`asm_execute`	Run an action or trigger you are building against a connection and return its output.	No
`asm_find_referencers`	Find the automations that use a given action, trigger or connection.	No
`asm_get_action_throttle`	Read the rate limit set on an action.	No
`asm_get_app_icon`	Check whether a connector, action or trigger has an icon, and return it.	No
`asm_get_connection_credentials`	Check whether a connection has stored credentials and list the credential names an action can use. Values are never returned.	No
`asm_get_connector_throttle`	Read the default rate limit set on a connector.	No
`asm_get_version`	Read one saved version of an action, trigger or connector.	No
`asm_get_webhook_handler`	Read the script a connector uses to route incoming webhooks to the right connected account.	No
`asm_inject_dynamic_fields`	Attach an existing field lookup to an action or trigger, so its fields come from the app's API.	No
`asm_inject_settings`	Set the configurable settings of an action or trigger, such as text boxes and dropdowns.	No
`asm_list_existing_operations`	List the actions and triggers already built on a connector.	No
`asm_list_versions`	List the saved versions of an action, trigger or connector, newest first.	No
`asm_load`	Load the full definition of an action, trigger or connector.	No
`asm_manage`	Finish an action, trigger or connector you built: check that it loads, and rename it or change its description.	No
`asm_read_assembly_logic`	Read the full logic of an action or trigger: its request, response mapping, fields, settings and pagination.	No
`asm_read_oauth_scopes`	Read the OAuth scopes declared on one of your connections.	No
`asm_replace_connector`	Move an action or trigger you built onto a different connector.	No
`asm_reset_settings`	Remove every setting from an action or trigger.	No
`asm_set_action_throttle`	Set or clear the rate limit on an action.	No
`asm_set_app_icon`	Set the icon of a connector, action or trigger from an image.	No
`asm_set_body`	Turn an action or trigger into a sequence of steps (requests, scripts, conditions, loops and calls to other actions) instead of a single request.	No
`asm_set_clear_encoding`	Declare how an app empties a field, so an automation can clear that field on purpose.	No
`asm_set_connector_throttle`	Set or clear the default rate limit on a connector.	No
`asm_set_field_catalog`	Let an action look up custom field IDs by name for each connected account when it runs.	No
`asm_set_output_fields`	Set the output fields an action or trigger returns.	No
`asm_set_pagination`	Set how an action or trigger pages through results, so one call collects every page.	No
`asm_set_polling_identity`	Set how a polling trigger recognizes records it has already seen.	No
`asm_set_require_one_of`	Declare a group of inputs of which at least one must be filled in.	No
`asm_set_required_scopes`	Declare the OAuth scopes an action or trigger needs.	No
`asm_set_response_errors`	Set how an action or trigger handles each kind of error response, such as retrying or failing.	No
`asm_set_webhook_handler`	Set or clear the script a connector uses to route incoming webhooks to the right connected account.	No
`asm_set_webhook_registration`	Make a trigger register its webhook with the app when an automation turns on, and remove it when the automation turns off.	No
`asm_set_writable_fields`	Record which fields from a field lookup the app accepts when writing.	No
`asm_test_api_connection`	Send a test request with a connection's credentials and report whether it succeeded.	No
`asm_test_api_endpoint`	Send an HTTP request, written as a curl command, with a connection's credentials and return the response.	No
`asm_test_webhook_handler`	Try a connector's webhook routing script on a sample payload without running any automation.	No

Error alerts (alert_)

Read and change the rules that rewrite error-alert messages for a step or an automation, read the system-wide rules, and manage dismissed alerts.

Rules for an automation or its steps can be changed by the automation's owner.
The retry and 401 lists apply to the whole installation and are read-only here.

Tool	What it does	In the assistant
`alert_get_automation_mappings`	Read the rules that rewrite error-alert messages for one automation.	No
`alert_get_error_policy`	Read which errors are retried and which app domains are exempt from turning an automation off after a 401 error.	No
`alert_get_mapping_trace`	Show, rule by rule, how an automation's alert rules rewrote the error from a given run.	No
`alert_get_step_mappings`	Read the alert-message rules on one step of an automation.	No
`alert_get_system_mappings`	Read the system-wide alert-message rules that apply to every automation.	No
`alert_list_suppressed`	List the alert messages that are dismissed for an automation.	No
`alert_set_automation_mappings`	Replace the alert-message rules for one automation, after showing a preview.	No
`alert_set_step_mappings`	Replace the alert-message rules on one step, after showing a preview.	No
`alert_suppress`	Dismiss an alert for an automation, so alerts with exactly that first line are no longer sent.	No
`alert_unsuppress`	Remove a dismissed alert, so that alert is sent again.	No

Publish (publish_)

Copy an automation, a folder of automations, or a connector, action or trigger from your dev environment to its prod environment.

Plan: Enterprise

Publishing works only from a dev environment, for what you own. Published copies land inactive in the prod environment's Automation Templates account. See Publish and deploy.

Tool	What it does	In the assistant
`publish_assembly`	Publish a connector, action or trigger you built from your dev environment to prod.	No
`publish_automation`	Publish an automation, with the subroutines and private connectors it depends on, from your dev environment to prod.	No
`publish_folder`	Publish every automation in a folder from your dev environment to prod, stopping at the first one that is rejected or skipped.	No

Credentials for your own app registration (credential_)

Report how Claude can obtain credentials for an app, and return the callback URL and scopes to register on the vendor's developer portal.

Tool	What it does	In the assistant
`credential_capabilities`	Report what your session can do to obtain credentials for an app, which decides how Claude goes about it.	No
`credential_registration_kit`	Return what you need to register your own app on a vendor's developer portal, including the callback URL and the scopes.	No

Other tools

Check the connection to the MCP server, report the installed plugin version, and file a bug report or feedback with APIANT.

Claude files a feedback report only after you approve the draft. Reports about the MCP server itself are filed without asking.

Tool	What it does	In the assistant
`deploy_get_plugin_version`	Report the version of the Claude Code plugin the server last saw for your install.	No
`ping`	Check that Claude is connected and signed in to APIANT, and report which account it is acting as.	Yes
`report_mcp_issue`	File a bug report or feedback with APIANT.	Yes

Next steps

Connect Claude Code to use these tools from your terminal.
Authentication and permissions explains the sign-in and the permissions the tools check.
Claude Code skills lists what you can ask Claude to do with them.
The APIANT assistant uses the tools marked above.

Related docs