fix(config): load discord.reply_to_mode from config.yaml into DISCORD_REPLY_TO_MODE

discord.reply_to_mode (and discord.extra.reply_to_mode) in config.yaml was
silently ignored — the gateway only honoured the DISCORD_REPLY_TO_MODE env
var, leaving the YAML setting with no effect.

Apply the same env-var bridge already used for require_mention, auto_thread,
free_response_channels, etc.  Also handles YAML 1.1 bare 'off' (parsed as
boolean False) correctly.

Top-level discord.reply_to_mode takes precedence over discord.extra.reply_to_mode;
a pre-existing DISCORD_REPLY_TO_MODE env var is never overwritten.

gateway/config.py

@@ -655,6 +655,17 @@ def load_gateway_config() -> GatewayConfig:
                     ):
                         if yaml_key in allow_mentions_cfg and not os.getenv(env_key):
                             os.environ[env_key] = str(allow_mentions_cfg[yaml_key]).lower()
+                # reply_to_mode: whether bot replies reference the triggering message (off/first/all)
+                # Supported at top-level (discord.reply_to_mode) or nested (discord.extra.reply_to_mode)
+                # Note: YAML 1.1 parses bare 'off' as boolean False, so we handle that explicitly.
+                _extra = discord_cfg.get("extra") if isinstance(discord_cfg.get("extra"), dict) else {}
+                _rtp_sources = (
+                    discord_cfg["reply_to_mode"] if "reply_to_mode" in discord_cfg
+                    else _extra.get("reply_to_mode")
+                )
+                if _rtp_sources is not None and not os.getenv("DISCORD_REPLY_TO_MODE"):
+                    _rtp_str = "off" if _rtp_sources is False else str(_rtp_sources).lower()
+                    os.environ["DISCORD_REPLY_TO_MODE"] = _rtp_str
 
             # Telegram settings → env vars (env vars take precedence)
             telegram_cfg = yaml_cfg.get("telegram", {})

tests/gateway/test_discord_reply_mode.py

@@ -15,7 +15,7 @@ from unittest.mock import MagicMock, AsyncMock, patch
 
 import pytest
 
-from gateway.config import PlatformConfig, GatewayConfig, Platform, _apply_env_overrides
+from gateway.config import PlatformConfig, GatewayConfig, Platform, _apply_env_overrides, load_gateway_config
 
 
 def _ensure_discord_mock():
@@ -396,3 +396,67 @@ class TestReplyToText:
         event = reply_text_adapter.handle_message.await_args.args[0]
         assert event.reply_to_message_id == "555"
         assert event.reply_to_text is None
+
+
+class TestYamlConfigLoading:
+    """Tests for reply_to_mode loaded from config.yaml discord section."""
+
+    def _write_config(self, tmp_path, content: str):
+        hermes_home = tmp_path / ".hermes"
+        hermes_home.mkdir()
+        (hermes_home / "config.yaml").write_text(content, encoding="utf-8")
+        return hermes_home
+
+    def test_top_level_reply_to_mode_off(self, tmp_path, monkeypatch):
+        """discord.reply_to_mode: off sets DISCORD_REPLY_TO_MODE env var."""
+        hermes_home = self._write_config(tmp_path, "discord:\n  reply_to_mode: off\n")
+        monkeypatch.setenv("HERMES_HOME", str(hermes_home))
+        monkeypatch.delenv("DISCORD_REPLY_TO_MODE", raising=False)
+
+        load_gateway_config()
+
+        assert os.environ.get("DISCORD_REPLY_TO_MODE") == "off"
+
+    def test_top_level_reply_to_mode_all(self, tmp_path, monkeypatch):
+        hermes_home = self._write_config(tmp_path, "discord:\n  reply_to_mode: all\n")
+        monkeypatch.setenv("HERMES_HOME", str(hermes_home))
+        monkeypatch.delenv("DISCORD_REPLY_TO_MODE", raising=False)
+
+        load_gateway_config()
+
+        assert os.environ.get("DISCORD_REPLY_TO_MODE") == "all"
+
+    def test_extra_reply_to_mode_off(self, tmp_path, monkeypatch):
+        """discord.extra.reply_to_mode: off is also honoured."""
+        hermes_home = self._write_config(
+            tmp_path, "discord:\n  extra:\n    reply_to_mode: \"off\"\n"
+        )
+        monkeypatch.setenv("HERMES_HOME", str(hermes_home))
+        monkeypatch.delenv("DISCORD_REPLY_TO_MODE", raising=False)
+
+        load_gateway_config()
+
+        assert os.environ.get("DISCORD_REPLY_TO_MODE") == "off"
+
+    def test_env_var_takes_precedence_over_yaml(self, tmp_path, monkeypatch):
+        """Existing DISCORD_REPLY_TO_MODE env var is not overwritten by YAML."""
+        hermes_home = self._write_config(tmp_path, "discord:\n  reply_to_mode: all\n")
+        monkeypatch.setenv("HERMES_HOME", str(hermes_home))
+        monkeypatch.setenv("DISCORD_REPLY_TO_MODE", "first")
+
+        load_gateway_config()
+
+        assert os.environ.get("DISCORD_REPLY_TO_MODE") == "first"
+
+    def test_top_level_takes_precedence_over_extra(self, tmp_path, monkeypatch):
+        """discord.reply_to_mode wins over discord.extra.reply_to_mode."""
+        hermes_home = self._write_config(
+            tmp_path,
+            "discord:\n  reply_to_mode: all\n  extra:\n    reply_to_mode: \"off\"\n",
+        )
+        monkeypatch.setenv("HERMES_HOME", str(hermes_home))
+        monkeypatch.delenv("DISCORD_REPLY_TO_MODE", raising=False)
+
+        load_gateway_config()
+
+        assert os.environ.get("DISCORD_REPLY_TO_MODE") == "all"