Tools

The Management MCP server exposes 40 tools, made up of 23 reads, 16 writes, and set_tenant_context, which selects the tenant a session works on. This page lists all of them so you can see exactly what an agent connected to your tenant is able to do. The tools are grouped here by toolset — the group names you can pass in the ?toolsets= connection parameter to offer an agent only part of the surface. A connection that names no toolsets gets every group. What an agent can actually do with a tool is still decided by your account’s permissions in the tenant, and by ?readOnly=true if you set it. Writes come in two kinds:

Write (additive) tools add something new and remove nothing.
Write (changes existing) tools can change, replace, or stop serving something that already exists. These are marked destructive to MCP clients, which is why some clients ask before running them. A tool is this kind if any call to it can change something that exists, even when other calls to it only add. configure_destination and the four upsert_* tools are examples.

There are no delete tools, by design. The server has no way to send a DELETE or a PATCH request — every call it can make is a read, or a create or update on a fixed allow-list checked when the server starts. No tool can delete a tenant, schema, environment, source, source group, domain, destination, client, or entity.

core

Always offered, whatever toolsets says: without these a session cannot say who it is or select a tenant.

Tool	Type	What it does
`whoami`	Read	Who the session is acting as, the tenants the account belongs to, the tenant currently selected, and how the connection’s parameters were read.
`set_tenant_context`	Tenant selection	Selects the tenant every later tool call acts on. Required on every signed-in session before any tenant-scoped tool, even for an account with one tenant. Changes nothing in the tenant.
`get_tenant_overview`	Read	Start here. The tenant’s plan, delivery regions, and environments — with the environment ids most other tools take.

schemas

The reason most people connect the server. The read tools let an agent design and prove a schema before anything is written.

Tool	Type	What it does
`list_mapping_schemas`	Read	Lists the tenant’s mapping schemas and where each is deployed. Pass `alias`, or `schemaId` and optionally `version`, to fetch one schema with its versions. Pass `query` to search schema source, for example to find which schema sets a property.
`get_schema_type_definitions`	Read	The type definitions an agent needs to write a valid schema.
`dry_run_mapping_schema`	Read	Runs a schema against input you supply and returns the output, without saving anything.
`dry_run_mapping_schema_against_entity`	Read	Same, against one of your real source entities.
`validate_mapping_schema_version`	Read	Checks whether a saved schema version compiles, without changing it.
`get_environment_deployments`	Read	What is live in an environment, and what is edited but not deployed yet.
`create_mapping_schema`	Write (additive)	Creates a new mapping schema and its first version. Deploys nothing.
`save_mapping_schema_version`	Write (changes existing)	Saves schema source into a version. An editable version is overwritten in place; a deployed one gets a new version instead.
`deploy_mapping_schema`	Write (changes existing)	Deploys schema versions to an environment, which changes what it serves. Always asks you to confirm in your client first.
`unpublish_mapping_schema`	Write (changes existing)	Removes a schema from one environment, so it stops serving that schema’s views there. The schema and its versions are untouched. Always asks you to confirm in your client first.

The dry-run and validate tools are reads, despite sounding like they change something. They work on a ?readOnly=true connection, so an agent can design, test, and validate a schema end to end without being able to save it — a good default even for authoring work, until you are happy with the result.

deploy_mapping_schema and unpublish_mapping_schema ask you to confirm in your client, by typing the environment’s name, before anything changes. Clients without MCP elicitation support are refused for these two tools. A client set to auto-approve can answer for you.

sources

Tool	Type	What it does
`list_sources`	Read	Lists sources and source groups, with their storage mode and the environments they publish to.
`list_source_entities`	Read	Browses or searches a source’s ingested entities. Pass `originId` to fetch one entity in full.
`upsert_source_group`	Write (changes existing)	Creates a source group. With `sourceGroup`, it changes an existing group’s name or type instead. A group’s alias and storage mode cannot be changed.
`upsert_source`	Write (changes existing)	Creates a source in a source group and attaches it to environments. With `source`, it changes an existing source’s name, or which environments it publishes to, instead. A source cannot move to another group.
`ingest_source_entities`	Write (changes existing)	Writes entities into a source. An entity with the same origin id is overwritten, and the schemas on that source reprocess. Always asks for confirmation first.

delivery

Tool	Type	What it does
`list_views`	Read	Lists what an environment has published. Pass `viewId` to fetch a single view’s content and the routes that point at it.
`inspect_route`	Read	Resolves a URL or handle and shows which route wins. The first stop for “why is this URL wrong?”.
`inspect_indexes`	Read	Lists indexes and their field definitions.
`get_index_document`	Read	Fetches a single indexed document.
`diagnose_missing_content`	Read	Answers “why is my content not showing up?”. Give it an `environment` and a source entity’s `originId`, a `url` that does not resolve, or both. Add `sourceId` if you know which source the entity is in. It checks the source, the entity, the deployed schema, the view, the route and the domain, and for a URL also whether it resolves and which environment clients can read it. It reports the first link that is broken and what to do about it. A check it could not complete is reported as `could_not_determine`, never as missing. It changes nothing.

domains

Tool	Type	What it does
`inspect_domains_and_hostnames`	Read	Domains and their hostnames, and optionally hostnames your content uses that no domain covers.
`upsert_domain`	Write (changes existing)	Creates a domain with one or more hostnames. With `domain`, it changes an existing domain’s name or the hostnames it covers instead.

destinations

Tool	Type	What it does
`list_destinations`	Read	Lists where the tenant’s published data is sent. Pass `alias` to read one destination and its configuration per environment.
`list_destination_apps`	Read	Lists the connectors a destination can be built on.
`create_destination`	Write (additive)	Creates a destination. It sends nothing until it is configured for an environment.
`configure_destination`	Write (changes existing)	Sets a destination up for one environment, or changes the setup that environment already has. It never removes a setting, but it can change a setting’s value, switch the setup off, change its regions, or point it at a different environment client. Credentials cannot be set through it, only in the Management App.

clients

Tool	Type	What it does
`list_environment_clients`	Read	Lists environment clients and their scopes. Access keys are never included.
`list_management_clients`	Read	Lists management clients. Access keys are never included.
`create_environment_client`	Write (additive)	Creates an environment client. The generated access key is not returned — fetch it in the Management App.
`update_environment_client`	Write (changes existing)	Changes an environment client’s name, scopes, index scopes, or allowed domains.
`rotate_management_client`	Write (changes existing)	Replaces a management client’s key. The previous key keeps working until a deadline the platform sets, which can be immediate. The new key is not returned.
`set_management_client_expiry`	Write (changes existing)	Sets or removes the moment a management client’s keys stop working.

On an environment client, an empty index scope means unrestricted, not no access. A partial update that dropped the field would therefore widen the client rather than narrow it. update_environment_client avoids this by reading the current client and merging your changes onto it, so an agent cannot silently remove an index restriction by omitting it — but it is worth knowing when you review what an agent changed.

observability

Tool	Type	What it does
`query_tenant_logs`	Read	Queries tenant logs — errors, warnings, processing activity.
`aggregate_tenant_logs`	Read	Counts log events grouped by field, for example errors by schema.
`get_tenant_metrics`	Read	Entity and index counts, request counts, and the processing backlog.

provisioning

Tool	Type	What it does
`create_tenant`	Write (additive)	Creates a new tenant, with you as its administrator. A tenant is billable, so this always asks for confirmation first.
`upsert_environment`	Write (changes existing)	Creates an empty environment. With `environment`, it renames that environment instead.

How the upsert tools decide what to do

upsert_environment, upsert_source_group, upsert_source, and upsert_domain each create a resource or change one. The arguments you pass decide which, never what already exists in your tenant.

Without the argument that names an existing resource (environment, sourceGroup, source, or domain), the call can only create. If a resource with that name already exists exactly as asked, nothing is written and the result says unchanged. If one exists with that name but differs, the call is refused and nothing changes. The refusal lists the differences and, where they can be changed, the call that would change them.
With that argument, the call only changes that one resource. If the argument does not match anything, the call is refused. It never falls back to creating something.

A misspelled edit therefore fails instead of creating something new. A call that succeeds reports an outcome of created, updated, or unchanged. Environment clients keep separate create_environment_client and update_environment_client tools, because an environment client is itself a credential. Creating one issues a key that can read your content, and updating one changes what that key can reach.

Tools that were renamed or merged

The tools below no longer exist. If a saved prompt, script, or log filter names one of them, use the call on the right instead. The arguments keep their names.

Earlier tool	Call instead
`get_mapping_schema`	`list_mapping_schemas` with `alias`, or with `schemaId` and optionally `version`
`search_mapping_schemas`	`list_mapping_schemas` with `query`
`get_source_entity`	`list_source_entities` with `sourceId` and `originId`
`get_view`	`list_views` with `viewId`
`get_destination`	`list_destinations` with `alias`
`create_environment`	`upsert_environment` with `name`
`update_environment`	`upsert_environment` with `environment` and `name`
`create_source_group`	`upsert_source_group` without `sourceGroup`
`update_source_group`	`upsert_source_group` with `sourceGroup`
`create_source`	`upsert_source` without `source`
`update_source`	`upsert_source` with `source`
`create_domain`	`upsert_domain` without `domain`
`update_domain`	`upsert_domain` with `domain`

Called with those arguments, each list_* tool returns what the earlier tool returned. The upsert tools report outcome where the earlier tools reported created or updated. Your tenant’s logs record the name of the tool that was actually called, so a log filter on an earlier tool name stops matching. Log entries for the upsert tools still carry an action named after the earlier tools, such as create_source or update_source.

What is deliberately not exposed

Some Management API capabilities are intentionally absent from the tool set:

Anything that deletes. No tenant, schema, environment, source, source group, domain, destination, client, view, or index can be removed through MCP.
Reading or regenerating a key on demand. No tool returns an access key, and the immediate key regeneration used when a key has leaked stays in the Management App. rotate_management_client is the routine rotation, with a grace window, and its new key is only readable in the app.
Bulk deploy paths. Deploying broadly stays in the Management App.

For the update tools, the permission is the verb, not the address. Several of the allowed updates share an address with a delete operation in the underlying API — the server can reach the update and has no way to reach the delete.

Next steps

Overview — hostname, authentication, connection parameters, guardrails, and client setup.
Query MCP — query your Enterspeed data instead of your configuration.

Overview

Schema reference

API

Integrations

Tutorials

Best practices

CLI

MCP server

core

schemas

sources

delivery

domains

destinations

clients

observability

provisioning

How the upsert tools decide what to do

Tools that were renamed or merged

What is deliberately not exposed

Next steps

​core

​schemas

​sources

​delivery

​domains

​destinations

​clients

​observability

​provisioning

​How the upsert tools decide what to do

​Tools that were renamed or merged

​What is deliberately not exposed

​Next steps

core

schemas

sources

delivery

domains

destinations

clients

observability

provisioning

How the upsert tools decide what to do

Tools that were renamed or merged

What is deliberately not exposed

Next steps