From 628ca99d9b2cdeb33f8bcbf31e77f281b4f283a3 Mon Sep 17 00:00:00 2001
From: Teknium <127238744+teknium1@users.noreply.github.com>
Date: Mon, 27 Apr 2026 08:43:24 -0700
Subject: [PATCH] fix(compression): show main + aux model and provider in
 feasibility warning (#16619)

The auto-lowered-threshold warning only named the compression model,
making it confusing when the main and aux models are configured with
the same slug but end up with different resolved context lengths (e.g.
OpenRouter's stepfun/step-3.5-flash catalog value vs. a main-model
context_length override). Users couldn't tell whether the warning
reflected two different models or a context-resolution mismatch.

Now includes both 'model (provider)' labels. The aux provider falls
back to the client's base_url hostname when the configured provider
is 'auto', so users see where compression is actually being called.
---
 run_agent.py | 46 ++++++++++++++++++++++++++++++++++++++++++----
 1 file changed, 42 insertions(+), 4 deletions(-)

diff --git a/run_agent.py b/run_agent.py
index 9801952352..85321628e6 100644
--- a/run_agent.py
+++ b/run_agent.py
@@ -2423,7 +2423,10 @@ class AIAgent:
         if not self.compression_enabled:
             return
         try:
-            from agent.auxiliary_client import get_text_auxiliary_client
+            from agent.auxiliary_client import (
+                _resolve_task_provider_model,
+                get_text_auxiliary_client,
+            )
             from agent.model_metadata import (
                 MINIMUM_CONTEXT_LENGTH,
                 get_model_context_length,
@@ -2433,6 +2436,14 @@ class AIAgent:
                 "compression",
                 main_runtime=self._current_main_runtime(),
             )
+            # Best-effort aux provider label for the warning message. The
+            # configured provider may be "auto", in which case we fall back
+            # to the client's base_url hostname so the user can still tell
+            # where the compression model is actually being called.
+            try:
+                _aux_cfg_provider, _, _, _, _ = _resolve_task_provider_model("compression")
+            except Exception:
+                _aux_cfg_provider = ""
             if client is None or not aux_model:
                 msg = (
                     "⚠ No auxiliary LLM provider configured — context "
@@ -2499,10 +2510,37 @@ class AIAgent:
                         new_threshold / main_ctx
                     )
                 safe_pct = int((aux_context / main_ctx) * 100) if main_ctx else 50
+                # Build human-readable "model (provider)" labels for both
+                # the main model and the compression model so users can
+                # tell at a glance which provider each side is actually
+                # using. When the configured provider is empty or "auto",
+                # fall back to the client's base_url hostname.
+                _main_model = getattr(self, "model", "") or "?"
+                _main_provider = getattr(self, "provider", "") or ""
+                _aux_provider_label = (
+                    _aux_cfg_provider
+                    if _aux_cfg_provider and _aux_cfg_provider != "auto"
+                    else ""
+                )
+                if not _aux_provider_label:
+                    try:
+                        from urllib.parse import urlparse
+                        _aux_provider_label = (
+                            urlparse(aux_base_url).hostname or aux_base_url
+                        )
+                    except Exception:
+                        _aux_provider_label = aux_base_url or "auto"
+                _main_label = (
+                    f"{_main_model} ({_main_provider})"
+                    if _main_provider
+                    else _main_model
+                )
+                _aux_label = f"{aux_model} ({_aux_provider_label})"
                 msg = (
-                    f"⚠ Compression model ({aux_model}) context is "
-                    f"{aux_context:,} tokens, but the main model's "
-                    f"compression threshold was {old_threshold:,} tokens. "
+                    f"⚠ Compression model {_aux_label} context is "
+                    f"{aux_context:,} tokens, but the main model "
+                    f"{_main_label}'s compression threshold was "
+                    f"{old_threshold:,} tokens. "
                     f"Auto-lowered this session's threshold to "
                     f"{new_threshold:,} tokens so compression can run.\n"
                     f"  To make this permanent, edit config.yaml — either:\n"