Show actual HTTP error in LLM test toast instead of 'empty response'

Track _last_error on every HF failure path (non-auth errors).
Test endpoint returns error field combining auth_error and last_error.
Toast now shows model name + real error e.g. 'HTTP 404: Model not found'

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

src/soci/api/routes.py

@@ -299,9 +299,11 @@ async def test_llm():
             user_message='Reply with exactly: {"ok": true}',
             max_tokens=32,
         )
-        return {"ok": bool(raw), "raw": raw, "provider": getattr(sim.llm, "provider", "?"),
+        error_detail = getattr(sim.llm, "_auth_error", "") or getattr(sim.llm, "_last_error", "")
+        return {"ok": bool(raw), "raw": raw,
+                "provider": getattr(sim.llm, "provider", "?"),
                 "model": getattr(sim.llm, "default_model", "?"),
-                "auth_error": getattr(sim.llm, "_auth_error", "")}
+                "error": error_detail}
     except Exception as e:
         return {"ok": False, "raw": "", "error": str(e)}
 

src/soci/engine/llm.py

@@ -845,6 +845,7 @@ class HFInferenceClient:
         )
         self._rate_limited_until: float = 0.0
         self._auth_error: str = ""
+        self._last_error: str = ""   # last non-auth error for diagnostics
 
     def _is_quota_exhausted(self) -> bool:
         return time.monotonic() < self._rate_limited_until
@@ -935,11 +936,13 @@ class HFInferenceClient:
                     logger.warning(f"HF model loading ({status}), waiting {wait:.0f}s")
                     await asyncio.sleep(wait)
                 else:
+                    self._last_error = f"HTTP {status}: {body}"
                     logger.error(f"HF HTTP error: {status} {body}")
                     if attempt == self.max_retries - 1:
                         return ""
                     await asyncio.sleep(2)
             except Exception as e:
+                self._last_error = str(e)
                 logger.error(f"HF error: {e}")
                 if attempt == self.max_retries - 1:
                     return ""

web/index.html

@@ -3413,8 +3413,8 @@ document.getElementById('llm-model').addEventListener('click', async (e) => {
         if (d.ok) {
           showToast(`✔ LLM OK — "${d.raw.slice(0,60)}"`, 'conv');
         } else {
-          const msg = d.auth_error ? `Auth error: ${d.auth_error.slice(0,80)}` : (d.error || d.raw || 'empty response');
-          showToast(`✘ LLM failed: ${msg}`, 'event');
+          const msg = (d.error || d.raw || 'empty response — model may not be on HF serverless API').slice(0, 120);
+          showToast(`✘ ${d.model}: ${msg}`, 'event');
         }
       } catch (err) { showToast('Test request failed', 'event'); }
     });