From 15e44527ab947a25e09e4fdb8ccbc8cf686eb26b Mon Sep 17 00:00:00 2001
From: teknium1 <127238744+teknium1@users.noreply.github.com>
Date: Tue, 30 Jun 2026 03:04:51 -0700
Subject: [PATCH] fix(copilot): prefer endpoints.api for base URL, guard empty
 chat base URL

Folds @trevorgordon981's #50590 into difujia's #15139:
- exchange_copilot_token now prefers the authoritative endpoints.api from
  the token-exchange response, falling back to the proxy-ep-derived host
- resolve_api_key_provider_credentials gains a copilot branch that resolves
  the account-specific base URL and a non-empty last-resort guard, so chat
  inference never wedges on an empty base URL (#50252)

Co-authored-by: Trevor Gordon <trevorbgordon@gmail.com>
---
 hermes_cli/auth.py         | 26 ++++++++++++++++++++++++++
 hermes_cli/copilot_auth.py | 29 +++++++++++++++++++++--------
 2 files changed, 47 insertions(+), 8 deletions(-)

diff --git a/hermes_cli/auth.py b/hermes_cli/auth.py
index 8ed80a777a8..d53f2cbbfc3 100644
--- a/hermes_cli/auth.py
+++ b/hermes_cli/auth.py
@@ -6357,6 +6357,26 @@ def resolve_api_key_provider_credentials(provider_id: str) -> Dict[str, Any]:
         base_url = _resolve_kimi_base_url(api_key, pconfig.inference_base_url, env_url)
     elif provider_id == "zai":
         base_url = _resolve_zai_base_url(api_key, pconfig.inference_base_url, env_url)
+    elif provider_id == "copilot":
+        # Resolve the Copilot API base URL from the token-exchange response
+        # (endpoints.api, with a proxy-ep fallback), which is authoritative
+        # for Enterprise / proxied accounts. Falls back to the registry
+        # default and is guarded non-empty below so chat inference never
+        # resolves an empty base URL (#50252).
+        base_url = env_url.rstrip("/") if env_url else pconfig.inference_base_url
+        try:
+            from hermes_cli.copilot_auth import (
+                resolve_copilot_token,
+                get_copilot_api_token,
+            )
+            raw_token, _ = resolve_copilot_token()
+            if raw_token:
+                _, resolved = get_copilot_api_token(raw_token)
+                resolved = (resolved or "").strip()
+                if resolved:
+                    base_url = resolved
+        except Exception as exc:
+            logger.debug("Copilot base URL resolution fell back to default: %s", exc)
     elif env_url:
         base_url = env_url.rstrip("/")
     else:
@@ -6365,6 +6385,12 @@ def resolve_api_key_provider_credentials(provider_id: str) -> Dict[str, Any]:
     if provider_id == "lmstudio":
         base_url = _normalize_lmstudio_runtime_base_url(base_url)
 
+    # Last-resort guard: an API-key provider must never hand back an empty
+    # base URL (a set-but-empty COPILOT_API_BASE_URL or similar env override
+    # otherwise wedges chat inference — #50252).
+    if not (isinstance(base_url, str) and base_url.strip()):
+        base_url = pconfig.inference_base_url
+
     return {
         "provider": provider_id,
         "api_key": api_key,
diff --git a/hermes_cli/copilot_auth.py b/hermes_cli/copilot_auth.py
index 69e472c8f85..1216254fea5 100644
--- a/hermes_cli/copilot_auth.py
+++ b/hermes_cli/copilot_auth.py
@@ -312,8 +312,10 @@ def exchange_copilot_token(raw_token: str, *, timeout: float = 10.0) -> tuple[st
 
     The returned token is a semicolon-separated string (not a standard JWT)
     used as ``Authorization: Bearer <token>`` for Copilot API requests.
-    If the token contains a ``proxy-ep`` field (enterprise accounts), the
-    derived API base URL is returned; otherwise ``base_url`` is None.
+    ``base_url`` is the account-specific API host: the authoritative
+    ``endpoints.api`` advertised by the exchange (enterprise/proxied
+    accounts), falling back to a host derived from the token's ``proxy-ep``
+    field. Individual accounts have neither, so ``base_url`` is None.
 
     Results are cached in-process and reused until close to expiry.
     Raises ``ValueError`` on failure.
@@ -354,10 +356,20 @@ def exchange_copilot_token(raw_token: str, *, timeout: float = 10.0) -> tuple[st
     # Convert expires_at to float if needed
     expires_at = float(expires_at) if expires_at else time.time() + 1800
 
-    # Derive enterprise base URL from proxy-ep in the token.
-    # The token is semicolon-separated: "tid=xxx;exp=xxx;proxy-ep=proxy.enterprise.githubcopilot.com;..."
-    # Replace leading "proxy." with "api." to get the API base URL.
-    base_url = _derive_base_url_from_proxy_ep(api_token)
+    # Resolve the account-specific API base URL. GitHub advertises the
+    # authoritative endpoint under ``endpoints.api`` in the exchange response
+    # (it differs for Copilot Enterprise / proxied accounts). When the
+    # response omits it, fall back to deriving the host from the ``proxy-ep``
+    # field embedded in the exchanged token. Individual accounts have neither,
+    # so ``base_url`` stays None and callers use the registry default.
+    base_url: Optional[str] = None
+    endpoints = data.get("endpoints")
+    if isinstance(endpoints, dict):
+        api_endpoint = str(endpoints.get("api") or "").strip().rstrip("/")
+        if api_endpoint:
+            base_url = api_endpoint
+    if not base_url:
+        base_url = _derive_base_url_from_proxy_ep(api_token)
 
     _jwt_cache[fp] = (api_token, expires_at, base_url)
     logger.debug(
@@ -408,8 +420,9 @@ def get_copilot_api_token(raw_token: str) -> tuple[str, Optional[str]]:
     account type). This preserves existing behaviour for accounts that don't
     need exchange while enabling access to internal-only models for those that do.
 
-    ``base_url`` is the enterprise API endpoint derived from the token's
-    ``proxy-ep`` field, or None for individual accounts.
+    ``base_url`` is the account-specific API endpoint advertised by the
+    exchange (``endpoints.api``, with a ``proxy-ep`` fallback), or None for
+    individual accounts.
     """
     if not raw_token:
         return raw_token, None