Fix AI Analyst: align model default, actionable provider-not-enabled error

- dashboard.py: change HF_MODEL default from Qwen2.5-3B to Qwen2.5-7B-Instruct-1M
(matches docker-compose ai_analyst service)
- Detect 400 model_not_supported / provider errors and return a clear message
with link to huggingface.co/settings/inference-providers
- /ai/debug: add "fix" field with provider setup instructions when 400 occurs
- docker-compose: add HF_TOKEN + HF_MODEL env vars to dashboard service
- index.html: error insights shown in red; status header turns red on error

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

dashboard/dashboard.py

@@ -28,7 +28,7 @@ FRONTEND_URL   = os.getenv("FRONTEND_URL",   "")
 
 # ── AI Analyst (inline LLM for on-demand generation) ───────────────────────────
 HF_TOKEN  = os.getenv("HF_TOKEN", "")
-HF_MODEL  = os.getenv("HF_MODEL", "Qwen/Qwen2.5-3B-Instruct")
+HF_MODEL  = os.getenv("HF_MODEL", "Qwen/Qwen2.5-7B-Instruct-1M")
 HF_URL    = "https://router.huggingface.co/v1/chat/completions"
 OLLAMA_HOST  = os.getenv("OLLAMA_HOST", "")
 OLLAMA_MODEL = os.getenv("OLLAMA_MODEL", "llama3.1:8b")
@@ -119,6 +119,13 @@ def _call_llm(prompt):
                 time.sleep(wait)
             else:
                 print(f"[Dashboard/LLM] HF error body: {r.text[:400]}")
+                try:
+                    err_code = r.json().get("error", {}).get("code", "")
+                except Exception:
+                    err_code = ""
+                if err_code == "model_not_supported" or "provider" in r.text.lower():
+                    return None, ("HF providers not enabled. Go to huggingface.co/settings/inference-providers "
+                                  "and enable at least one provider (e.g. Cerebras, Groq, or HF Inference).")
                 return None, f"HF HTTP {r.status_code}: {r.text[:120]}"
         except requests.exceptions.Timeout:
             print(f"[Dashboard/LLM] HF timeout (attempt {attempt+1})")
@@ -607,7 +614,13 @@ def ai_debug():
         result["http_status"] = r.status_code
         result["response_body"] = r.text[:500]
         try:
-            result["response_json"] = r.json()
+            rj = r.json()
+            result["response_json"] = rj
+            err_code = rj.get("error", {}).get("code", "")
+            if err_code == "model_not_supported" or "provider" in r.text.lower():
+                result["fix"] = ("Enable inference providers at: "
+                                 "https://huggingface.co/settings/inference-providers  "
+                                 "Enable at least one free provider: Cerebras, Groq, SambaNova, or HF Inference.")
         except Exception:
             pass
     except Exception as e:

dashboard/templates/index.html

@@ -467,7 +467,10 @@
       const ph   = document.getElementById("ai-placeholder");
       if (ph) ph.remove();
       const div  = document.createElement("div");
+      const isErr = insight.source === "error" || insight.source === "config"
+                    || (insight.text || "").startsWith("⚠️");
       div.className = "insight-card insight-new";
+      if (isErr) div.style.cssText = "border-left-color:#e53935; background:#fff5f5;";
       const t = new Date(insight.timestamp * 1000).toLocaleTimeString();
       div.innerHTML = `<div class="insight-time">${t}</div><div>${insight.text}</div>`;
       list.prepend(div);
@@ -1128,8 +1131,11 @@
       eventSource.addEventListener("ai_insight", (e) => {
         const insight = JSON.parse(e.data);
         addInsight(insight);
-        document.getElementById("ai-status").textContent =
-          "Last update: " + new Date().toLocaleTimeString();
+        const isErr = insight.source === "error" || insight.source === "config"
+                      || (insight.text || "").startsWith("⚠️");
+        const statusEl = document.getElementById("ai-status");
+        statusEl.textContent = isErr ? "error — see below" : "Last update: " + new Date().toLocaleTimeString();
+        statusEl.style.color = isErr ? "#e53935" : "#999";
         const btn = document.getElementById("ai-generate-btn");
         if (btn._tick) { clearInterval(btn._tick); btn._tick = null; }
         btn.disabled = false;

docker-compose.yml

@@ -179,6 +179,8 @@ services:
     environment:
       - MATCHER_URL=http://matcher:6000
       - FRONTEND_URL=http://localhost:5000
+      - HF_TOKEN=${HF_TOKEN:-}
+      - HF_MODEL=${HF_MODEL:-Qwen/Qwen2.5-7B-Instruct-1M}
 
 volumes:
   matcher_data:  # Persists SQLite database across container restarts