- Published on
Claude Tag API Enterprise Beta β Control @Claude Per Slack Channel and Attribute Every API Request to an Individual User
If your whole team is using @Claude in Slack, how does an admin control what it can do in each channel? Until now, the answer was organization-level settings in the Anthropic console. The Claude Tag API brings that control down to the channel level β and lets you manage it through code.
Anthropic released two enterprise API betas across October 20β21, 2026.
Claude Tag API: Manage Per-Channel @Claude Settings Programmatically
Claude Tag is Anthropic''s Slack integration that responds when someone types @Claude in a workspace. Many teams use it as a personal productivity tool, but for enterprises, "which channels can use AI responses, and at what level" becomes a compliance question.
Using the Claude Tag API beta header (experimental-claude-tag-2026-10-21), you can interact with the /v1/organizations/claude_tag/ endpoints to:
- Read and update channel-level and workspace-level settings: Change how Claude Tag behaves in a specific channel β response mode, allowed features, and more β through API calls.
- Inspect setting inheritance: See which settings are applied to a given channel, and whether they come from the organization default or a channel-specific override.
- Attach Access Bundles: Link specific Claude feature bundles to individual channels, controlling exactly which capabilities are available where.
For IT admins, this is Slack channel policy automation. Instead of manually configuring Claude Tag settings each time a new project channel is created, you can include a Claude Tag API call in your channel provisioning script β consistent policy, applied automatically from day one.
User Profiles API: Tie API Requests to Individual End Users
The second beta is the User Profiles API. This is designed for platforms, gateways, and resellers β companies that use the Anthropic API to serve their own end customers.
Previously, all API requests were treated as coming from the same organization. If one end user violated Anthropic''s usage policies, the enforcement action could affect the entire organization''s API access. The User Profiles API solves this.
POST /v1/user_profiles
β Create a profile for each end user
Request header:
anthropic-user-profile-id: <profile_id>
β Attribute this request to a specific end user
The key outcome: safety enforcement can be applied at the individual user level rather than the entire organization. If a specific user''s account shows repeated policy violations, only that user profile is restricted β the rest of the platform''s users are unaffected.
According to Releasebot, profiles now also display enforcement status and action history, with a filter to find restricted or banned profiles.
What These Two APIs Signal Together
The Claude Tag API and User Profiles API are separate features, but they point in the same direction: AI governance expressed as code.
As AI usage scales across an organization, questions like "who is using AI, in which channels, with what permissions" become harder to answer manually. Until now, the answer required navigating Anthropic''s admin UI and working with organization-level policies. These two APIs bring that control into the developer''s codebase.
How to Access the Beta
Both APIs are available on Claude Enterprise plans and require including the appropriate beta header in each request.
- Claude Tag API:
experimental-claude-tag-2026-10-21header - User Profiles API:
user-profiles-2026-09-04header (earlier versions:user-profiles-2026-08-18,user-profiles-2026-03-24)
The correct request fields differ depending on which beta header version you use. Check Anthropic''s official API reference to confirm the right parameters for your header version before integrating.
An EdTech Take on These Two APIs
For EdTech platforms using the Claude API, the User Profiles API has meaningful implications.
Create a profile per student account, and if one student''s usage triggers a policy action, only that student''s profile is affected β the rest of the class continues learning without interruption. The Claude Tag API similarly opens up the possibility of differentiated @Claude configurations: advanced analysis for teacher channels, constrained responses in student channels.
Both features represent a maturation of AI tooling toward "enterprise-grade governance" β the kind of control that institutional buyers increasingly require before rolling out AI across their organizations.
Source: