fix(auth): self-heal Codex refresh_token rotation by reimporting from ~/.codex

Hermes keeps its own copy of the Codex OAuth token per profile and at the
top level, separate from the Codex CLI's ~/.codex/auth.json. OAuth
refresh_tokens are single-use, so when the Codex CLI (or another Hermes
process) rotates the shared token, the frozen copy's refresh_token goes
stale and refresh_codex_oauth_pure fails with a relogin-required error
(invalid_grant / refresh_token_reused / 401). Today that surfaces as a hard
401 on the turn — idle profiles and desktop sessions 401 "token_expired"
until a manual re-auth — even though ~/.codex/auth.json holds a fresh token.

_refresh_codex_auth_tokens now falls back to _import_codex_cli_tokens() (the
canonical Codex CLI store) when the stored refresh_token is rejected, adopts
and persists the fresh token, and lets the in-flight retry succeed. This
complements PR #6525 (force relogin on 401/403): we attempt automatic
recovery before surfacing a relogin prompt. Transient failures (e.g. 429
quota, relogin_required=False) are never self-healed — the stored token is
still valid there — so they re-raise unchanged, and the happy path is
untouched.

Adds tests/hermes_cli/test_auth_codex_self_heal.py covering: self-heal on
invalid_grant, no self-heal on 429 quota, re-raise when ~/.codex is absent,
and happy-path-unchanged.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

hermes_cli/auth.py

@@ -3660,11 +3660,37 @@ def _refresh_codex_auth_tokens(
     
     Saves the new tokens to Hermes auth store automatically.
     """
-    refreshed = refresh_codex_oauth_pure(
+    try:
-        str(tokens.get("access_token", "") or ""),
+        refreshed = refresh_codex_oauth_pure(
-        str(tokens.get("refresh_token", "") or ""),
+            str(tokens.get("access_token", "") or ""),
-        timeout_seconds=timeout_seconds,
+            str(tokens.get("refresh_token", "") or ""),
-    )
+            timeout_seconds=timeout_seconds,
+        )
+    except AuthError as exc:
+        # Self-heal cross-store refresh_token rotation. Hermes keeps its OWN
+        # Codex OAuth token (per profile + top-level), separate from the Codex
+        # CLI's ~/.codex/auth.json. OAuth refresh_tokens are single-use, so when
+        # the Codex CLI (or another Hermes process) rotates the shared token,
+        # this frozen copy's refresh_token goes stale and the refresh fails with
+        # a relogin-required error (invalid_grant / refresh_token_reused / 401).
+        # Before surfacing that as a hard 401 to the turn, adopt the canonical
+        # fresh token from ~/.codex/auth.json (the Codex CLI keeps it current) so
+        # idle profiles / desktop sessions recover automatically instead of
+        # 401'ing until a manual re-auth. Transient failures (e.g. 429 quota)
+        # keep relogin_required=False — the stored token is still valid there, so
+        # we never self-heal those and re-raise unchanged.
+        if not getattr(exc, "relogin_required", False):
+            raise
+        imported = _import_codex_cli_tokens()
+        if not (imported and str(imported.get("access_token", "") or "").strip()):
+            raise
+        logger.info(
+            "Codex refresh_token rejected (%s); recovered from ~/.codex/auth.json.",
+            getattr(exc, "code", None) or "auth_error",
+        )
+        _save_codex_tokens(imported)
+        return dict(imported)
+
     updated_tokens = dict(tokens)
     updated_tokens["access_token"] = refreshed["access_token"]
     updated_tokens["refresh_token"] = refreshed["refresh_token"]

tests/hermes_cli/test_auth_codex_self_heal.py

@@ -0,0 +1,120 @@
+"""Regression tests for Codex refresh_token self-heal (cross-store rotation).
+
+Hermes keeps its OWN copy of the Codex OAuth token (per profile + top-level),
+separate from the Codex CLI's ``~/.codex/auth.json``. OAuth refresh_tokens are
+single-use, so when the Codex CLI (or another Hermes process) rotates the shared
+token, the frozen copy's refresh_token goes stale and ``refresh_codex_oauth_pure``
+fails with a relogin-required error. ``_refresh_codex_auth_tokens`` must then
+recover by re-importing the canonical token from ``~/.codex/auth.json`` instead of
+surfacing a hard 401 — but ONLY for relogin-required failures, never for transient
+ones (e.g. 429 quota, where the stored token is still valid).
+"""
+
+import pytest
+
+import hermes_cli.auth as auth
+from hermes_cli.auth import AuthError, _refresh_codex_auth_tokens
+
+STALE = {"access_token": "stale-access", "refresh_token": "stale-refresh"}
+
+
+def test_self_heals_on_stale_refresh_token(monkeypatch):
+    """invalid_grant (relogin-required) → reimport from ~/.codex and persist it."""
+    saved = {}
+    fresh = {
+        "access_token": "fresh-access",
+        "refresh_token": "fresh-refresh",
+        "last_refresh": "2026-06-12T00:00:00Z",
+    }
+
+    def _rejected(*_a, **_k):
+        raise AuthError(
+            "refresh token rejected",
+            provider="openai-codex",
+            code="invalid_grant",
+            relogin_required=True,
+        )
+
+    monkeypatch.setattr(auth, "refresh_codex_oauth_pure", _rejected)
+    monkeypatch.setattr(auth, "_import_codex_cli_tokens", lambda: dict(fresh))
+    monkeypatch.setattr(auth, "_save_codex_tokens", lambda t, *a, **k: saved.update(t))
+
+    out = _refresh_codex_auth_tokens(STALE, 20.0)
+
+    assert out["access_token"] == "fresh-access"
+    assert out["refresh_token"] == "fresh-refresh"
+    # the recovered token was persisted to the Hermes auth store
+    assert saved["access_token"] == "fresh-access"
+
+
+def test_does_not_self_heal_on_rate_limit(monkeypatch):
+    """429 quota keeps relogin_required=False — token still valid, must NOT reimport."""
+    import_calls = {"n": 0}
+
+    def _rate_limited(*_a, **_k):
+        raise AuthError(
+            "quota exhausted",
+            provider="openai-codex",
+            code="codex_rate_limited",
+            relogin_required=False,
+        )
+
+    def _import_spy():
+        import_calls["n"] += 1
+        return {"access_token": "should-not-be-used"}
+
+    monkeypatch.setattr(auth, "refresh_codex_oauth_pure", _rate_limited)
+    monkeypatch.setattr(auth, "_import_codex_cli_tokens", _import_spy)
+    monkeypatch.setattr(auth, "_save_codex_tokens", lambda *a, **k: None)
+
+    with pytest.raises(AuthError) as ei:
+        _refresh_codex_auth_tokens(STALE, 20.0)
+
+    assert ei.value.code == "codex_rate_limited"
+    assert import_calls["n"] == 0  # never touched ~/.codex on a transient failure
+
+
+def test_reraises_when_codex_cli_token_absent(monkeypatch):
+    """relogin-required but ~/.codex unavailable/expired → propagate original error."""
+
+    def _reused(*_a, **_k):
+        raise AuthError(
+            "refresh token reused",
+            provider="openai-codex",
+            code="refresh_token_reused",
+            relogin_required=True,
+        )
+
+    monkeypatch.setattr(auth, "refresh_codex_oauth_pure", _reused)
+    monkeypatch.setattr(auth, "_import_codex_cli_tokens", lambda: None)
+    monkeypatch.setattr(auth, "_save_codex_tokens", lambda *a, **k: None)
+
+    with pytest.raises(AuthError) as ei:
+        _refresh_codex_auth_tokens(STALE, 20.0)
+
+    assert ei.value.code == "refresh_token_reused"
+
+
+def test_happy_path_unchanged(monkeypatch):
+    """Normal refresh succeeds → rotated tokens persisted, ~/.codex never consulted."""
+    saved = {}
+    import_calls = {"n": 0}
+
+    def _import_spy():
+        import_calls["n"] += 1
+        return None
+
+    monkeypatch.setattr(
+        auth,
+        "refresh_codex_oauth_pure",
+        lambda *a, **k: {"access_token": "rotated", "refresh_token": "rotated-r"},
+    )
+    monkeypatch.setattr(auth, "_import_codex_cli_tokens", _import_spy)
+    monkeypatch.setattr(auth, "_save_codex_tokens", lambda t, *a, **k: saved.update(t))
+
+    out = _refresh_codex_auth_tokens({"access_token": "a", "refresh_token": "b"}, 20.0)
+
+    assert out["access_token"] == "rotated"
+    assert out["refresh_token"] == "rotated-r"
+    assert saved["access_token"] == "rotated"
+    assert import_calls["n"] == 0  # happy path must not consult ~/.codex