common/hermes
2
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity

fix(gateway): make reasoning session-scoped by default

Browse Source
This commit is contained in:
Iris Jin
2026-04-25 02:39:01 -04:00
committed by Teknium
parent 489bed6f96
commit 25ba6a4a74
4 changed files with 269 additions and 23 deletions
Download Patch File Download Diff File Expand all files Collapse all files

134
tests/gateway/test_reasoning_command.py
View File

@@ -33,6 +33,7 @@ def _make_runner():
runner._ephemeral_system_prompt = ""
runner._prefill_messages = []
runner._reasoning_config = None
runner._session_reasoning_overrides = {}
runner._show_reasoning = False
runner._provider_routing = {}
runner._fallback_model = None
@@ -76,6 +77,10 @@ class TestReasoningCommand:
source = inspect.getsource(gateway_run.GatewayRunner._handle_message)
assert '"reasoning"' in source
def test_parse_reasoning_command_args_accepts_ascii_and_smart_global_flags(self):
assert gateway_run.GatewayRunner._parse_reasoning_command_args("high --global") == ("high", True)
assert gateway_run.GatewayRunner._parse_reasoning_command_args("—global xhigh") == ("xhigh", True)
@pytest.mark.asyncio
async def test_reasoning_command_reloads_current_state_from_config(self, tmp_path, monkeypatch):
hermes_home = tmp_path / "hermes"
@@ -111,13 +116,90 @@ class TestReasoningCommand:
runner = _make_runner()
runner._reasoning_config = {"enabled": True, "effort": "medium"}
result = await runner._handle_reasoning_command(_make_event("/reasoning low"))
result = await runner._handle_reasoning_command(_make_event("/reasoning low --global"))
saved = yaml.safe_load(config_path.read_text(encoding="utf-8"))
assert saved["agent"]["reasoning_effort"] == "low"
assert runner._reasoning_config == {"enabled": True, "effort": "low"}
assert "takes effect on next message" in result
@pytest.mark.asyncio
async def test_handle_reasoning_command_defaults_to_session_only(self, tmp_path, monkeypatch):
hermes_home = tmp_path / "hermes"
hermes_home.mkdir()
config_path = hermes_home / "config.yaml"
config_path.write_text("agent:\n reasoning_effort: medium\n", encoding="utf-8")
monkeypatch.setattr(gateway_run, "_hermes_home", hermes_home)
runner = _make_runner()
event = _make_event("/reasoning high")
session_key = runner._session_key_for_source(event.source)
result = await runner._handle_reasoning_command(event)
saved = yaml.safe_load(config_path.read_text(encoding="utf-8"))
assert saved["agent"]["reasoning_effort"] == "medium"
assert runner._session_reasoning_overrides[session_key] == {"enabled": True, "effort": "high"}
assert runner._reasoning_config == {"enabled": True, "effort": "high"}
assert "session only" in result
@pytest.mark.asyncio
async def test_reasoning_global_clears_existing_session_override(self, tmp_path, monkeypatch):
hermes_home = tmp_path / "hermes"
hermes_home.mkdir()
config_path = hermes_home / "config.yaml"
config_path.write_text("agent:\n reasoning_effort: medium\n", encoding="utf-8")
monkeypatch.setattr(gateway_run, "_hermes_home", hermes_home)
runner = _make_runner()
event = _make_event("/reasoning low --global")
session_key = runner._session_key_for_source(event.source)
runner._session_reasoning_overrides[session_key] = {"enabled": True, "effort": "xhigh"}
result = await runner._handle_reasoning_command(event)
saved = yaml.safe_load(config_path.read_text(encoding="utf-8"))
assert saved["agent"]["reasoning_effort"] == "low"
assert session_key not in runner._session_reasoning_overrides
assert "saved to config" in result
@pytest.mark.asyncio
async def test_reasoning_reset_clears_session_override_without_config_write(self, tmp_path, monkeypatch):
hermes_home = tmp_path / "hermes"
hermes_home.mkdir()
config_path = hermes_home / "config.yaml"
config_path.write_text("agent:\n reasoning_effort: medium\n", encoding="utf-8")
monkeypatch.setattr(gateway_run, "_hermes_home", hermes_home)
runner = _make_runner()
event = _make_event("/reasoning reset")
session_key = runner._session_key_for_source(event.source)
runner._session_reasoning_overrides[session_key] = {"enabled": True, "effort": "xhigh"}
result = await runner._handle_reasoning_command(event)
saved = yaml.safe_load(config_path.read_text(encoding="utf-8"))
assert saved["agent"]["reasoning_effort"] == "medium"
assert session_key not in runner._session_reasoning_overrides
assert "cleared" in result
def test_resolve_session_reasoning_prefers_session_override(self, tmp_path, monkeypatch):
hermes_home = tmp_path / "hermes"
hermes_home.mkdir()
(hermes_home / "config.yaml").write_text("agent:\n reasoning_effort: low\n", encoding="utf-8")
monkeypatch.setattr(gateway_run, "_hermes_home", hermes_home)
runner = _make_runner()
source = _make_event("/reasoning").source
session_key = runner._session_key_for_source(source)
runner._session_reasoning_overrides[session_key] = {"enabled": True, "effort": "xhigh"}
assert runner._resolve_session_reasoning_config(source=source) == {"enabled": True, "effort": "xhigh"}
def test_run_agent_reloads_reasoning_config_per_message(self, tmp_path, monkeypatch):
hermes_home = tmp_path / "hermes"
hermes_home.mkdir()
@@ -167,6 +249,56 @@ class TestReasoningCommand:
assert _CapturingAgent.last_init is not None
assert _CapturingAgent.last_init["reasoning_config"] == {"enabled": True, "effort": "low"}
def test_run_agent_prefers_session_reasoning_override(self, tmp_path, monkeypatch):
hermes_home = tmp_path / "hermes"
hermes_home.mkdir()
(hermes_home / "config.yaml").write_text("agent:\n reasoning_effort: low\n", encoding="utf-8")
monkeypatch.setattr(gateway_run, "_hermes_home", hermes_home)
monkeypatch.setattr(gateway_run, "_env_path", hermes_home / ".env")
monkeypatch.setattr(gateway_run, "load_dotenv", lambda *args, **kwargs: None)
monkeypatch.setattr(
gateway_run,
"_resolve_runtime_agent_kwargs",
lambda: {
"provider": "openrouter",
"api_mode": "chat_completions",
"base_url": "https://openrouter.ai/api/v1",
"api_key": "***",
},
)
fake_run_agent = types.ModuleType("run_agent")
fake_run_agent.AIAgent = _CapturingAgent
monkeypatch.setitem(sys.modules, "run_agent", fake_run_agent)
_CapturingAgent.last_init = None
runner = _make_runner()
session_key = "agent:main:local:dm"
runner._session_reasoning_overrides[session_key] = {"enabled": True, "effort": "high"}
source = SessionSource(
platform=Platform.LOCAL,
chat_id="cli",
chat_name="CLI",
chat_type="dm",
user_id="user-1",
)
result = asyncio.run(
runner._run_agent(
message="ping",
context_prompt="",
history=[],
source=source,
session_id="session-1",
session_key=session_key,
)
)
assert result["final_response"] == "ok"
assert _CapturingAgent.last_init is not None
assert _CapturingAgent.last_init["reasoning_config"] == {"enabled": True, "effort": "high"}
def test_run_agent_includes_enabled_mcp_servers_in_gateway_toolsets(self, tmp_path, monkeypatch):
hermes_home = tmp_path / "hermes"
hermes_home.mkdir()

5
tests/gateway/test_session_model_override_routing.py
View File

@@ -54,6 +54,7 @@ def _make_runner():
runner._background_tasks = set()
runner._session_db = None
runner._session_model_overrides = {}
runner._session_reasoning_overrides = {}
runner._pending_model_notes = {}
runner._pending_approvals = {}
runner._agent_cache = {}
@@ -102,6 +103,7 @@ def test_run_agent_prefers_session_override_over_global_runtime(monkeypatch):
)
session_key = "agent:main:local:dm"
runner._session_model_overrides[session_key] = _codex_override()
runner._session_reasoning_overrides[session_key] = {"enabled": True, "effort": "high"}
result = asyncio.run(
runner._run_agent(
@@ -121,6 +123,7 @@ def test_run_agent_prefers_session_override_over_global_runtime(monkeypatch):
assert _CapturingAgent.last_init["api_mode"] == "codex_responses"
assert _CapturingAgent.last_init["base_url"] == "https://chatgpt.com/backend-api/codex"
assert _CapturingAgent.last_init["api_key"] == "***"
assert _CapturingAgent.last_init["reasoning_config"] == {"enabled": True, "effort": "high"}
@pytest.mark.asyncio
@@ -149,6 +152,7 @@ async def test_background_task_prefers_session_override_over_global_runtime(monk
)
session_key = runner._session_key_for_source(source)
runner._session_model_overrides[session_key] = _codex_override()
runner._session_reasoning_overrides[session_key] = {"enabled": True, "effort": "high"}
await runner._run_background_task("say hello", source, "bg_test")
@@ -158,3 +162,4 @@ async def test_background_task_prefers_session_override_over_global_runtime(monk
assert _CapturingAgent.last_init["api_mode"] == "codex_responses"
assert _CapturingAgent.last_init["base_url"] == "https://chatgpt.com/backend-api/codex"
assert _CapturingAgent.last_init["api_key"] == "***"
assert _CapturingAgent.last_init["reasoning_config"] == {"enabled": True, "effort": "high"}

15
tests/gateway/test_session_model_reset.py
View File

@@ -1,4 +1,4 @@
"""Tests that /new (and its /reset alias) clears the session-scoped model override."""
"""Tests that /new (and its /reset alias) clears session-scoped overrides."""
from datetime import datetime
from types import SimpleNamespace
from unittest.mock import AsyncMock, MagicMock
@@ -37,6 +37,7 @@ def _make_runner():
runner._voice_mode = {}
runner.hooks = SimpleNamespace(emit=AsyncMock(), loaded_hooks=False)
runner._session_model_overrides = {}
runner._session_reasoning_overrides = {}
runner._pending_model_notes = {}
runner._background_tasks = set()
@@ -75,14 +76,16 @@ async def test_new_command_clears_session_model_override():
runner._session_model_overrides[session_key] = {
"model": "gpt-4o",
"provider": "openai",
"api_key": "sk-test",
"api_key": "***",
"base_url": "",
"api_mode": "openai",
}
runner._session_reasoning_overrides[session_key] = {"enabled": True, "effort": "high"}
await runner._handle_reset_command(_make_event("/new"))
assert session_key not in runner._session_model_overrides
assert session_key not in runner._session_reasoning_overrides
@pytest.mark.asyncio
@@ -92,10 +95,12 @@ async def test_new_command_no_override_is_noop():
session_key = build_session_key(_make_source())
assert session_key not in runner._session_model_overrides
assert session_key not in runner._session_reasoning_overrides
await runner._handle_reset_command(_make_event("/new"))
assert session_key not in runner._session_model_overrides
assert session_key not in runner._session_reasoning_overrides
@pytest.mark.asyncio
@@ -115,12 +120,16 @@ async def test_new_command_only_clears_own_session():
runner._session_model_overrides[other_key] = {
"model": "claude-sonnet-4-6",
"provider": "anthropic",
"api_key": "sk-ant-test",
"api_key": "***",
"base_url": "",
"api_mode": "anthropic",
}
runner._session_reasoning_overrides[session_key] = {"enabled": True, "effort": "high"}
runner._session_reasoning_overrides[other_key] = {"enabled": True, "effort": "low"}
await runner._handle_reset_command(_make_event("/new"))
assert session_key not in runner._session_model_overrides
assert other_key in runner._session_model_overrides
assert session_key not in runner._session_reasoning_overrides
assert other_key in runner._session_reasoning_overrides