#32677 [Bug]: DB/API-registered MCP servers don't resolve os.environ/ in stdio env (same root cause as #31050, different field)
### Check for existing issues
- [x] I have searched the existing issues and checked that my issue is not a duplicate.
### What happened?
MCP servers created via `POST /v1/mcp/server` (DB-backed) never resolve `os.environ/VAR` in the stdio `env` map. `config.yaml` MCP servers resolve `os.environ/` at load via `ProxyConfig._check_for_os_environ_vars()`, but the DB builder `build_mcp_server_from_table()` deserializes `env` verbatim, so the literal `os.environ/...` string is passed to the subprocess and auth fails. This is the stdio-`env` sibling of #31050 (which covers http `static_headers`); nothing currently tracks `env`.
**Source** (verified at v1.89.0; unchanged through v1.91.0 / v1.92.0-rc.1): - `litellm/proxy/proxy_server.py` â `_check_for_os_environ_vars` (L3571-3606) resolves `os.environ/` recursively over the whole config dict, invoked at L3766; `mcp_servers` is loaded afterward (L4679-4690), so YAML entries are already resolved. - `litellm/proxy/_experimental/mcp_server/mcp_server_manager.py` â `build_mcp_server_from_table` (L1005+) JSON-deserializes `env` (L1013) and hands it to `MCPServer(env=...)` (L1148) with no `get_secret` / `_check_for_os_environ_vars` call. The fâŚ