Claude Enterprise

The Claude Enterprise Source connects an AI System to a Claude Enterprise organization on claude.ai. Modulos reads the activity feed, the directory, the settings, the plugins, and the adoption and cost data of that organization. Modulos only reads. It does not change or delete data in the Claude organization.

Used by	Metric Tests, Agent Tests, Scout, Risk Agent
Mode	Pull
Credentials	A Compliance Access Key, an Analytics API key, or the two keys
Metric selection	Type the metric name

For the Claude API organization on platform.claude.com, use the Claude Console Source. It uses a different key.

What this is

The Claude Enterprise Source reads six items. Each item needs one key with one scope:

Item	Key	Scope
Activity feed	Compliance Access Key	read:compliance_activities
Organizations, roles and groups	Compliance Access Key	read:compliance_org_data
Users and group members	Compliance Access Key	read:compliance_user_data
Organization settings	Compliance Access Key	read:compliance_org_data
Plugins and marketplaces	Compliance Access Key	read:compliance_org_data
Adoption and cost	Analytics API key	read:analytics

You can use one key or the two keys. Tests and AI agents read only the items that your keys give access to.

Where in Modulos

• AI System → Settings → Sources → Add Source → Claude Enterprise adds the Source.
• AI System → Settings → Sources → select the Source shows What this source can read to the users who can open the Settings of the AI System.

Who can do what

The Sources page has the general permissions. For this Source, the Owners of the AI System can open What this source can read and select Check again. Editors and Viewers cannot open the Settings of the AI System and do not see the Source there.

Before you start

• An Organization Admin makes Claude Enterprise available. See Source & Connectors.
• The primary owner of the Claude organization activates the Compliance API in claude.ai, in Organization settings → API.

How to create the keys

Create the keys in claude.ai, in Organization settings → API:

1. In the Keys section, create a Compliance Access Key.
2. Select the scopes read:compliance_activities, read:compliance_org_data, and read:compliance_user_data.
3. Do not select delete:compliance_user_data.
4. If Modulos must read adoption and cost data, create an Analytics API key with the scope read:analytics.

A Compliance Access Key starts with sk-ant-api01-. You cannot add a scope to a key after you create it.

How to connect Claude Enterprise

1. Open AI System → Settings → Sources.
2. Select Add Source.
3. Select Claude Enterprise.
4. Type a Source Name.
5. If you have a Compliance Access Key, paste it into Compliance Access Key.
6. If you have an Analytics API key, paste it into Analytics API Key.
7. Keep API URL (optional) empty. Modulos then uses the API of Anthropic.
8. Select Save.

You must enter one or two keys.

123

Add a Claude Enterprise Source.

1. 1
   Compliance Access Key

   Reads activity, organizations, users, settings, and plugins.
2. 2
   Analytics API Key

   Reads seats, active users, usage, and cost.
3. 3
   API URL

   Keep it empty. Modulos then uses the API of Anthropic.

What this source can read

When you save a Claude Enterprise Source, Modulos checks what each key can read. If Modulos can read no item, it does not save the Source and shows the cause, for example that the Compliance API is not active. Open the Source to see What this source can read, with one status for each item. In AI System → Settings → Sources, the Connection column shows Healthy when Modulos can read one or more items.

Status	What to do
Available	Tests and AI agents can read the item.
No key	Add the key that the item needs.
Key rejected	Anthropic did not accept the key. Create a new key and enter it again.
Wrong key type	Create the correct type of key in claude.ai in Organization settings → API.
Missing scope	Create a new key with the scope. You cannot change the scopes of a key.
Not enabled	The primary owner activates the Compliance API. For Adoption and cost, Anthropic must activate the Analytics API. If only Organization settings is not enabled, Anthropic must activate it for your parent organization. For these two cases, speak to your Anthropic representative.
Unknown	Modulos could not do the check. Select Check again after some time.

After a change in claude.ai that keeps the same keys, select Check again. For example, do this after the primary owner activates the Compliance API. For a new key, edit the Source and enter the key. If an active Compliance Access Key of the Claude organization has the scope delete:compliance_user_data, the panel shows a warning with the name of the key.

123

What this source can read, with one status for each item.

1. 1
   Available

   Tests and AI agents can read the item.
2. 2
   Not enabled

   The item is not active. The text under it gives the cause.
3. 3
   Check again

   Tests the keys again and shows the new statuses.

Metrics for Metric Tests

A Metric Test on a Claude Enterprise Source reads one of these metrics:

1. In the test dialog, type the name as the table shows it.
2. Select Check.

A metric gives a value only when its item is Available. The metrics of Plugins and marketplaces and of Organization settings also must have Organizations, roles and groups. If an item is not available, Check does not find its metrics.

Metric	Value	Item
assigned_seats	Assigned seats on the most recent day of the analytics data	Adoption and cost
active_users_7d	Users who used Claude (chat, Claude Code, or Cowork) in the last 7 days. Each user counts one time.	Adoption and cost
active_users_30d	The same count for the last 30 days	Adoption and cost
seat_utilization_30d	Part of the assigned seats, from 0 to 1, that have a holder who used Claude in the last 30 days	Adoption and cost
cost_per_active_user_30d_usd	Spend in USD for the last 30 days, divided by active_users_30d. Only for Enterprise plans that pay by usage. Plans that pay by seat report usage credits.	Adoption and cost
mcp_server_changes_30d	MCP servers (connectors) added, changed, or removed in the last 30 days	Activity feed
inference_hook_denials_7d	Requests that the inference hooks of the organization blocked in the last 7 days	Activity feed
data_exports_30d	Exports of the audit log, the data, or the member list in the last 30 days	Activity feed
attributed_report_views_30d	Views and downloads of Smart Reports that attribute usage findings to individual members in the last 30 days, sessions opened from such a report, and administrators shown which members a recommendation suggests	Activity feed
library_submissions_30d	Skills or plugins that members submitted to the organization library in the last 30 days	Activity feed
plugin_version_changes_30d	Changes to the plugin version that members get in the last 30 days	Activity feed
member_owned_plugins	Plugins in the personal marketplaces of members	Plugins and marketplaces
remote_reach_plugins	Plugins that declare an MCP server or a CLI. Plugins that Anthropic did not classify are not counted.	Plugins and marketplaces
plugins_not_passing_scan	Plugins with a current version that has a content scan result other than "pass"	Plugins and marketplaces
sso_enforced	1 if single sign-on is enforced in each linked organization that controls the setting, 0 if not. No value if no linked organization controls the setting.	Organization settings
chat_retention_days	The longest chat retention of the linked organizations, in days. No limit, or no period set by an administrator, counts as 36,500 days.	Organization settings
content_redaction_enabled	1 if content redaction is on in each linked organization that controls the setting, 0 if not. No value if no linked organization controls the setting.	Organization settings
ip_allowlist_configured	1 if an IP allowlist with one or more ranges is on in each linked organization that controls the setting, 0 if not. No value if no linked organization controls the setting.	Organization settings

If the cost data does not include all 30 days, cost_per_active_user_30d_usd gives no value.

Scout, Agent Tests, and the Risk Agent

Scout, Agent Tests, and the Risk Agent read the items of a Claude Enterprise Source that are Available:

• Activity feed: sign-ins, plugin events, exports, blocked requests, and changes to settings, connectors, and MCP servers.
• Organizations, roles and groups: organizations, custom roles, and groups.
• Users and group members: members with their roles.
• Organization settings: the settings in effect, for example single sign-on, content redaction, IP allowlist, and retention, and the Compliance Access Keys.
• Plugins and marketplaces: the plugins in the organization and personal marketplaces, with their "remote reach" and their content scan result.
• Adoption and cost: seats, invites, spend, the active users of chat, Claude Code, and Cowork, and the use of connectors, skills, and plugins.

If Modulos can read no item, Scout tells you why. See Scout, Agent Tests, and Risk Agent.

Important considerations

• Give each key the read scopes only. Modulos does not write to the Claude organization.
• After you save, the keys are hidden. When you edit the Source, enter again each key that the Source must keep. An empty key field removes the stored key.

Related pages

Sources

All Source types and shared settings

Claude Console

The Source for a Claude API organization on platform.claude.com

Tests & Schedules

Metric Tests and their schedules

Agent Tests

AI agents that check your Sources