Restricting a custom endpoint to specific users — is there a supported approach today?

[MPeli]

Hi,

We are running LibreChat with a custom endpoint connecting to an AI agent deployed on AWS Bedrock AgentCore. The endpoint uses the user's Entra (Azure AD) OIDC token, forwarded via the {{LIBRECHAT_OPENID_TOKEN}} header placeholder:

custom:
  - name: "ET RTB chatbot"
    baseURL: "https://bedrock-agentcore..."
    directEndpoint: true
    headers:
      Authorization: "Bearer {{LIBRECHAT_OPENID_TOKEN}}"
      X-Amzn-Bedrock-AgentCore-Runtime-Custom-SessionId: "{{LIBRECHAT_BODY_CONVERSATIONID}}"
    models:
      default: ["model_a", "model_b"]
      fetch: true

We need to restrict access to this endpoint to a specific subset of users, not all authenticated users.

From reviewing the docs and the codebase, it appears there is currently no native way to do this — endpoints.custom has no allowedUsers, allowedRoles, or allowedGroups field. The only workarounds we see are:

1. Enforce authorization on the backend — the OIDC token is already forwarded, so Bedrock AgentCore can reject unauthorized users. This works but the UX is poor: the endpoint is still visible to everyone, they only get an error when they try to use it.
2. OPENID_REQUIRED_ROLE — gates the entire application, not a single endpoint.

We also looked at migrating to the Agents endpoint to leverage its native ACL/sharing system, but endpoints.agents has no headers: field and no documented mechanism for forwarding {{LIBRECHAT_OPENID_TOKEN}} to the underlying backend. So that path would break our per-user token forwarding.

We noticed the open enhancement #3137 — Admin panel that allows easy user monitoring/UI configuration per user and the 2026 roadmap, which describes the upcoming v1 Admin Panel with:

Configuration Profiles — Apply custom configuration profiles to roles initially, with group-level profiles as a stretch goal
Access Controls Throughout — Access controls embedded into every aspect of the admin panel

This sounds like it would eventually solve our use case.

Our questions:

1. Is our analysis correct — is there currently no supported way to restrict a custom endpoint to specific users?
2. Is there a recommended workaround we may have missed?
3. Will the Agents endpoint ever support {{LIBRECHAT_OPENID_TOKEN}} (or equivalent) header forwarding, so that migrating to Agents + ACL becomes viable for setups like ours?
4. Will the upcoming Admin Panel / Configuration Profiles (Enhancement: Admin panel that allows easy user monitoring/UI configuration per user #3137) cover per-endpoint access control for custom endpoints specifically?

Thanks for the great work on the roadmap — the Admin Panel looks like it will make a real difference for enterprise deployments.

[danny-avila]

Yes, this will be covered by the admin panel

[MPeli]

Thank you!

[LeoCME]

Hi @MPeli ,

We are considering connecting LibreChat to AWS AgentCore code interpreter as well leveraging AgentCore MCP. Is the setup working well for you? Do you see any drawbacks compared to LibreChat native code interpreter ?