From 60391d0eef1fb96b6c3ea94531576874d67b09b1 Mon Sep 17 00:00:00 2001
From: AIalliAI <285906080+AIalliAI@users.noreply.github.com>
Date: Thu, 11 Jun 2026 19:50:06 +0000
Subject: [PATCH] fix(agent): don't apply Codex gpt-5.5 autoraise notice when
 an external context engine is active
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

When context.engine selects a plugin engine (e.g. LCM), the host
compression threshold — including the Codex gpt-5.5 50% -> 85%
autoraise — only configures the built-in ContextCompressor and never
reaches the plugin. The autoraise notice still fired, telling the user
auto-compaction was raised when nothing actually changed, and the
startup context-limit line printed the host percent next to the
engine's own threshold_tokens, contradicting itself.

- Clear _compression_threshold_autoraised when a plugin engine is
  selected, suppressing both the CLI startup notice and the gateway
  turn-1 replay via _compression_warning.
- Print the active engine's own threshold_percent in the startup
  context-limit line so percent and token count agree.
- Built-in behavior is preserved, including the fallback path where a
  configured engine fails to load and the built-in compressor takes
  over.

Fixes #44439
---
 agent/agent_init.py                           | 14 ++-
 .../test_plugin_context_engine_init.py        | 98 +++++++++++++++++++
 2 files changed, 111 insertions(+), 1 deletion(-)

diff --git a/agent/agent_init.py b/agent/agent_init.py
index 36bea43f44b..32a06259441 100644
--- a/agent/agent_init.py
+++ b/agent/agent_init.py
@@ -1773,6 +1773,12 @@ def init_agent(
 
     if _selected_engine is not None:
         agent.context_compressor = _selected_engine
+        # External engines own compaction policy: the host compression
+        # threshold (including the Codex gpt-5.5 autoraise above) only
+        # configures the built-in ContextCompressor and never reaches the
+        # plugin, so the autoraise notice would announce a change that does
+        # not apply. Drop it. (#44439)
+        agent._compression_threshold_autoraised = None
         # Resolve context_length for plugin engines — mirrors switch_model() path
         from agent.model_metadata import get_model_context_length
         _plugin_ctx_len = get_model_context_length(
@@ -2012,7 +2018,13 @@ def init_agent(
 
     if not agent.quiet_mode:
         if compression_enabled:
-            print(f"📊 Context limit: {agent.context_compressor.context_length:,} tokens (compress at {int(compression_threshold*100)}% = {agent.context_compressor.threshold_tokens:,})")
+            # Report the active engine's own threshold — for a plugin engine
+            # the host compression_threshold is not in effect, and mixing the
+            # two printed a percent that contradicted the token count. (#44439)
+            _active_threshold_pct = getattr(
+                agent.context_compressor, "threshold_percent", compression_threshold
+            )
+            print(f"📊 Context limit: {agent.context_compressor.context_length:,} tokens (compress at {int(_active_threshold_pct*100)}% = {agent.context_compressor.threshold_tokens:,})")
         else:
             print(f"📊 Context limit: {agent.context_compressor.context_length:,} tokens (auto-compression disabled)")
         # Notice with the exact opt-back-out command. Printed inline at startup
diff --git a/tests/run_agent/test_plugin_context_engine_init.py b/tests/run_agent/test_plugin_context_engine_init.py
index 7285cb1f625..f4b29ef0093 100644
--- a/tests/run_agent/test_plugin_context_engine_init.py
+++ b/tests/run_agent/test_plugin_context_engine_init.py
@@ -139,3 +139,101 @@ def test_plugin_engine_update_model_args():
     assert "model" in kw
     assert "provider" in kw
     assert "api_mode" in kw
+
+
+def _codex_agent_kwargs():
+    return dict(
+        model="gpt-5.5",
+        provider="openai-codex",
+        api_key="test-key-1234567890",
+        base_url="https://chatgpt.com/backend-api/codex",
+        quiet_mode=True,
+        skip_context_files=True,
+        skip_memory=True,
+    )
+
+
+def test_codex_gpt55_autoraise_suppressed_for_plugin_engine():
+    """Codex gpt-5.5 autoraise must not fire when an external engine is active.
+
+    Regression test for #44439 — the host compression threshold (including
+    the 0.85 autoraise) never reaches a plugin context engine, so the notice
+    announced a change that did not apply.
+    """
+    engine = _StubEngine()
+    cfg = {
+        "context": {"engine": "stub"},
+        "compression": {"enabled": True, "threshold": 0.75},
+        "agent": {},
+    }
+
+    with (
+        patch("hermes_cli.config.load_config", return_value=cfg),
+        patch("plugins.context_engine.load_context_engine", return_value=engine),
+        patch("agent.model_metadata.get_model_context_length", return_value=272_000),
+        patch("run_agent.get_tool_definitions", return_value=[]),
+        patch("run_agent.check_toolset_requirements", return_value={}),
+        patch("run_agent.OpenAI"),
+    ):
+        from run_agent import AIAgent
+
+        agent = AIAgent(**_codex_agent_kwargs())
+
+    assert agent.context_compressor is engine
+    assert agent._compression_threshold_autoraised is None
+    assert agent._compression_warning is None
+    # The engine's own policy is untouched by the host threshold.
+    assert engine.threshold_percent == 0.75
+
+
+def test_codex_gpt55_autoraise_still_applies_to_builtin_compressor():
+    """Stock built-in compressor keeps the 50% → 85% Codex gpt-5.5 autoraise."""
+    cfg = {
+        "compression": {"enabled": True, "threshold": 0.50},
+        "agent": {},
+    }
+
+    with (
+        patch("hermes_cli.config.load_config", return_value=cfg),
+        patch("agent.context_compressor.get_model_context_length", return_value=272_000),
+        patch("run_agent.get_tool_definitions", return_value=[]),
+        patch("run_agent.check_toolset_requirements", return_value={}),
+        patch("run_agent.OpenAI"),
+    ):
+        from run_agent import AIAgent
+
+        agent = AIAgent(**_codex_agent_kwargs())
+
+    assert agent._compression_threshold_autoraised == {"from": 0.50, "to": 0.85}
+    assert agent.context_compressor.threshold_percent == 0.85
+    # Gateway parity: the notice is stashed for replay on turn 1.
+    assert agent._compression_warning and "85%" in agent._compression_warning
+
+
+def test_codex_gpt55_autoraise_applies_when_plugin_engine_missing():
+    """If the configured engine fails to load, the built-in compressor is
+    active and the autoraise (plus its notice) must still apply."""
+    cfg = {
+        "context": {"engine": "no-such-engine"},
+        "compression": {"enabled": True, "threshold": 0.50},
+        "agent": {},
+    }
+
+    with (
+        patch("hermes_cli.config.load_config", return_value=cfg),
+        patch(
+            "plugins.context_engine.load_context_engine",
+            side_effect=ValueError("not found"),
+        ),
+        patch("hermes_cli.plugins.get_plugin_context_engine", return_value=None),
+        patch("agent.context_compressor.get_model_context_length", return_value=272_000),
+        patch("run_agent.get_tool_definitions", return_value=[]),
+        patch("run_agent.check_toolset_requirements", return_value={}),
+        patch("run_agent.OpenAI"),
+    ):
+        from run_agent import AIAgent
+
+        agent = AIAgent(**_codex_agent_kwargs())
+
+    assert agent._compression_threshold_autoraised == {"from": 0.50, "to": 0.85}
+    assert agent.context_compressor.threshold_percent == 0.85