## What happened?

On the **Virtual Keys** UI page (and `GET /key/list` directly), filtering by **User ID** or **Key Alias** returns *all* keys instead of narrowing, whenever the caller has team-wide key visibility — e.g. a regular team member whose team grants `/key/list` in `team_member_permissions`, or a team admin. On our deployment (one shared team, ~3,300 members, ~4,040 keys) the filters have effectively never worked: every filtered query returns all 4,040 keys, byte-identical to the unfiltered response.

### Root cause

In `_build_key_filter_conditions` (`litellm/proxy/management_endpoints/key_management_endpoints.py`):

- `user_id` / `key_alias` are only applied inside the caller's **"own keys" OR-branch** (`user_condition`, [L5465–L5489](https://github.com/BerriAI/litellm/blob/88e03e548716a45284597edf2b7f47a7e6a66d5f/litellm/proxy/management_endpoints/key_management_endpoints.py#L5465-L5489)). - Team-wide visibility is added as a **separate, unfiltered OR-branch**: `{"team_id": {"in": admin_team_ids}}` ([L5519–L5520](https://github.com/BerriAI/litellm/blob/88e03e548716a45284597edf2b7f47a7e6a66d5f/litellm/proxy/management_endpoints/key_management_endpoints.py#L5519-L5520))…