mirror of
https://github.com/NousResearch/hermes-agent.git
synced 2026-07-24 16:54:43 +00:00
fa4800414c
Follow-ups on top of the cherry-picked #62644 mechanism, porting it to current main and closing the salvage-review requirements: - proactive_prune_min_reclaim_tokens (default 4096): a prune only COMMITS when it reclaims a meaningful token batch, measured on the pruned output. A committed prune rewrites already-sent history and invalidates the provider prompt-cache prefix; this hysteresis gate keeps those breaks episodic/amortized (like a compression boundary) instead of firing every tool iteration. 0 disables the gate. (Design point credited to the #62389 review cycle's prune_minimum_tokens.) - Standard no-op caller contract: every skip path returns the INPUT list object; the loop commits only on 'result is not messages' + non-zero count. - Loop call is getattr+callable guarded (plugin engines predating the hook, SimpleNamespace test doubles) and exception-swallowed at debug level. - Config parse follows the compression.max_attempts hardened semantics: booleans rejected, fractional floats rejected, integral floats/numeric strings accepted; negative trigger = disabled. - cli-config.yaml.example documented (all three keys) and gateway _CACHE_BUSTING_CONFIG_KEYS extended so hot-reload rebuilds the agent. - Tests: min-reclaim gate both directions, input-object no-op contract, no-orphan tool_call_id pairing in BOTH directions (#69830 pin rule), default-off zero-behavior-change pin, config parse seam, and behavioral loop-wiring tests (consulted/commit/no-op/absent-method/raising).
111 lines
4.3 KiB
Python
111 lines
4.3 KiB
Python
"""compression.proactive_prune_* — config parse seam for the proactive prune.
|
|
|
|
Mirrors ``test_compression_max_attempts_config.py``: the three knobs are
|
|
parsed in ``agent_init`` with the same hardened semantics (booleans rejected,
|
|
fractional floats rejected — not truncated, integral floats and numeric
|
|
strings accepted) and attached to the built-in compressor. Default is
|
|
0 / 8000 / 4096, i.e. the feature is OFF and behavior-neutral unless
|
|
``proactive_prune_tokens`` is set above 0.
|
|
"""
|
|
|
|
from __future__ import annotations
|
|
|
|
import contextlib
|
|
import io
|
|
from pathlib import Path
|
|
|
|
from hermes_state import SessionDB
|
|
from run_agent import AIAgent
|
|
|
|
|
|
def _config(**prune_keys) -> dict:
|
|
compression = {
|
|
"enabled": True,
|
|
"threshold": 0.50,
|
|
"target_ratio": 0.20,
|
|
"protect_first_n": 3,
|
|
"protect_last_n": 20,
|
|
}
|
|
compression.update(prune_keys)
|
|
return {
|
|
"compression": compression,
|
|
"prompt_caching": {"cache_ttl": "5m"},
|
|
"sessions": {},
|
|
"bedrock": {},
|
|
}
|
|
|
|
|
|
def _make_agent(monkeypatch, tmp_path: Path, **prune_keys):
|
|
from hermes_cli import config as config_mod
|
|
|
|
monkeypatch.setattr(config_mod, "load_config", lambda: _config(**prune_keys))
|
|
db = SessionDB(db_path=tmp_path / "state.db")
|
|
with contextlib.redirect_stdout(io.StringIO()):
|
|
agent = AIAgent(
|
|
base_url="https://chatgpt.com/backend-api/codex",
|
|
api_key="test-key",
|
|
provider="openai-codex",
|
|
model="gpt-5.5",
|
|
enabled_toolsets=[],
|
|
disabled_toolsets=[],
|
|
quiet_mode=True,
|
|
skip_memory=True,
|
|
session_db=db,
|
|
session_id="proactive-prune-config-test",
|
|
)
|
|
return agent
|
|
|
|
|
|
class TestProactivePruneConfig:
|
|
def test_default_is_disabled_when_unset(self, monkeypatch, tmp_path):
|
|
agent = _make_agent(monkeypatch, tmp_path)
|
|
cc = agent.context_compressor
|
|
assert cc.proactive_prune_tokens == 0
|
|
assert cc.proactive_prune_min_result_chars == 8000
|
|
assert cc.proactive_prune_min_reclaim_tokens == 4096
|
|
|
|
def test_custom_values_are_honored(self, monkeypatch, tmp_path):
|
|
agent = _make_agent(
|
|
monkeypatch,
|
|
tmp_path,
|
|
proactive_prune_tokens=48_000,
|
|
proactive_prune_min_result_chars=12_000,
|
|
proactive_prune_min_reclaim_tokens=8_192,
|
|
)
|
|
cc = agent.context_compressor
|
|
assert cc.proactive_prune_tokens == 48_000
|
|
assert cc.proactive_prune_min_result_chars == 12_000
|
|
assert cc.proactive_prune_min_reclaim_tokens == 8_192
|
|
|
|
def test_boolean_is_rejected_not_coerced(self, monkeypatch, tmp_path):
|
|
# bool subclasses int: YAML `proactive_prune_tokens: true` must fall
|
|
# back to disabled, never coerce to 1 token.
|
|
agent = _make_agent(monkeypatch, tmp_path, proactive_prune_tokens=True)
|
|
assert agent.context_compressor.proactive_prune_tokens == 0
|
|
|
|
def test_fractional_float_is_rejected_not_truncated(self, monkeypatch, tmp_path):
|
|
agent = _make_agent(monkeypatch, tmp_path, proactive_prune_tokens=48_000.7)
|
|
assert agent.context_compressor.proactive_prune_tokens == 0
|
|
|
|
def test_integral_float_and_numeric_string_accepted(self, monkeypatch, tmp_path):
|
|
agent = _make_agent(monkeypatch, tmp_path, proactive_prune_tokens=48_000.0)
|
|
assert agent.context_compressor.proactive_prune_tokens == 48_000
|
|
agent = _make_agent(monkeypatch, tmp_path, proactive_prune_tokens="32000")
|
|
assert agent.context_compressor.proactive_prune_tokens == 32_000
|
|
|
|
def test_negative_trigger_treated_as_disabled(self, monkeypatch, tmp_path):
|
|
agent = _make_agent(monkeypatch, tmp_path, proactive_prune_tokens=-100)
|
|
assert agent.context_compressor.proactive_prune_tokens == 0
|
|
|
|
def test_garbage_falls_back_to_defaults(self, monkeypatch, tmp_path):
|
|
agent = _make_agent(
|
|
monkeypatch,
|
|
tmp_path,
|
|
proactive_prune_tokens="lots",
|
|
proactive_prune_min_result_chars=None,
|
|
proactive_prune_min_reclaim_tokens="???",
|
|
)
|
|
cc = agent.context_compressor
|
|
assert cc.proactive_prune_tokens == 0
|
|
assert cc.proactive_prune_min_result_chars == 8000
|
|
assert cc.proactive_prune_min_reclaim_tokens == 4096
|