mirror of
https://github.com/NousResearch/hermes-agent.git
synced 2026-06-23 10:42:00 +00:00
04730f32e7
Adds hermes_cli/context_switch_guard.py mirroring the model_cost_guard pattern. When a user switches models mid-session (Herm TUI picker, CLI, or /model on Telegram/Discord), the warning surfaces on the existing ModelSwitchResult.warning_message path used by the expensive-model guard if the new model's compression threshold is below the current session size. Partial fix for #23767 — addresses only the 'user-facing guardrail when switching from a high-context provider to a substantially lower-context provider' slice. The other proposed fixes from that issue (hard preflight token guard, metadata cache invalidation on switch, compression safety invariant, oversized tool-output handling) are out of scope for this PR.
105 lines
No EOL
3.1 KiB
Python
105 lines
No EOL
3.1 KiB
Python
"""Tests for hermes_cli.context_switch_guard."""
|
|
|
|
from __future__ import annotations
|
|
|
|
from types import SimpleNamespace
|
|
|
|
from hermes_cli.context_switch_guard import merge_preflight_compression_warning
|
|
from hermes_cli.model_switch import ModelSwitchResult
|
|
|
|
|
|
def _result(*, model: str = "small-model") -> ModelSwitchResult:
|
|
return ModelSwitchResult(
|
|
success=True,
|
|
new_model=model,
|
|
target_provider="openrouter",
|
|
provider_changed=False,
|
|
api_key="k",
|
|
base_url="https://example.com/v1",
|
|
api_mode="chat_completions",
|
|
provider_label="openrouter",
|
|
model_info={"context_length": 32_000},
|
|
)
|
|
|
|
|
|
def _compressor(monkeypatch, *, context_length: int = 200_000):
|
|
from agent.context_compressor import ContextCompressor
|
|
|
|
monkeypatch.setattr(
|
|
"agent.context_compressor.get_model_context_length",
|
|
lambda *a, **k: context_length,
|
|
)
|
|
return ContextCompressor(
|
|
model="big-model",
|
|
threshold_percent=0.5,
|
|
protect_first_n=3,
|
|
protect_last_n=20,
|
|
quiet_mode=True,
|
|
config_context_length=context_length,
|
|
)
|
|
|
|
|
|
def test_no_warning_when_below_new_threshold(monkeypatch):
|
|
monkeypatch.setattr(
|
|
"hermes_cli.context_switch_guard.resolve_display_context_length",
|
|
lambda *a, **k: 32_000,
|
|
)
|
|
cc = _compressor(monkeypatch)
|
|
cc.last_prompt_tokens = 10_000
|
|
agent = SimpleNamespace(
|
|
context_compressor=cc,
|
|
compression_enabled=True,
|
|
conversation_history=[],
|
|
base_url="",
|
|
api_key="",
|
|
)
|
|
result = _result()
|
|
merge_preflight_compression_warning(result, agent=agent)
|
|
assert not result.warning_message
|
|
|
|
|
|
def test_warns_when_estimate_exceeds_new_threshold(monkeypatch):
|
|
monkeypatch.setattr(
|
|
"hermes_cli.context_switch_guard.resolve_display_context_length",
|
|
lambda *a, **k: 32_000,
|
|
)
|
|
monkeypatch.setattr(
|
|
"hermes_cli.context_switch_guard._estimate_tokens",
|
|
lambda *a, **k: 90_000,
|
|
)
|
|
cc = _compressor(monkeypatch)
|
|
agent = SimpleNamespace(
|
|
context_compressor=cc,
|
|
compression_enabled=True,
|
|
conversation_history=[],
|
|
base_url="",
|
|
api_key="",
|
|
)
|
|
result = _result()
|
|
merge_preflight_compression_warning(result, agent=agent)
|
|
assert result.warning_message
|
|
assert "preflight compression" in result.warning_message
|
|
assert "shrinks" in result.warning_message
|
|
|
|
|
|
def test_merge_appends_to_existing_warning(monkeypatch):
|
|
monkeypatch.setattr(
|
|
"hermes_cli.context_switch_guard._estimate_tokens",
|
|
lambda *a, **k: 90_000,
|
|
)
|
|
monkeypatch.setattr(
|
|
"hermes_cli.context_switch_guard.resolve_display_context_length",
|
|
lambda *a, **k: 32_000,
|
|
)
|
|
cc = _compressor(monkeypatch)
|
|
agent = SimpleNamespace(
|
|
context_compressor=cc,
|
|
compression_enabled=True,
|
|
base_url="",
|
|
api_key="",
|
|
)
|
|
result = _result()
|
|
result.warning_message = "expensive"
|
|
merge_preflight_compression_warning(result, agent=agent)
|
|
assert "expensive" in result.warning_message
|
|
assert "preflight compression" in result.warning_message |