Update core code, compliance, ui, legal emails

Dockerfile

@@ -31,23 +31,17 @@ RUN npm install -g openclaw@latest \
     && (command -v openclaw || true)
 
 # Install Ollama AS ROOT - pinned version, force amd64
-# Using pinned version URL to avoid redirect issues during Docker build
-# Mark as OPTIONAL - app works without it via HF Inference API fallback
+# REQUIRED - build fails if download fails or binary is corrupt
 RUN echo "Downloading Ollama v0.15.2 (amd64)..." && \
     curl -fSL --retry 3 --retry-delay 10 --max-time 300 \
         "https://github.com/ollama/ollama/releases/download/v0.15.2/ollama-linux-amd64" \
         -o /usr/local/bin/ollama && \
     chmod +x /usr/local/bin/ollama && \
-    OLLAMA_SIZE=$(stat -c%s /usr/local/bin/ollama 2>/dev/null || echo 0) && \
     echo "Ollama binary: $(file /usr/local/bin/ollama)" && \
-    echo "Size: ${OLLAMA_SIZE} bytes" && \
-    if [ "$OLLAMA_SIZE" -lt 1000000 ]; then \
-        echo "WARNING: Ollama binary too small (${OLLAMA_SIZE} bytes) — likely truncated download"; \
-        rm -f /usr/local/bin/ollama; \
-    else \
-        echo "OK: Ollama binary looks good"; \
-    fi \
-    || echo "WARNING: Ollama download failed - will use HF Inference API only"
+    echo "Size: $(stat -c%s /usr/local/bin/ollama) bytes" && \
+    file /usr/local/bin/ollama | grep -q "ELF" || (echo "FATAL: Ollama binary is not ELF" && exit 1) && \
+    test $(stat -c%s /usr/local/bin/ollama) -gt 1000000 || (echo "FATAL: Ollama binary too small" && exit 1) && \
+    echo "OK: Ollama verified"
 
 # Create HF-required user (uid 1000)
 RUN useradd -m -u 1000 user

app.py

@@ -160,7 +160,9 @@ def generate_with_ollama(model: str, prompt: str) -> Optional[str]:
 
 
 def generate_with_hf_api(prompt: str, model: str = None) -> Optional[str]:
-    """Generate text using HuggingFace Inference API (free, no GPU needed)."""
+    """Generate text using HuggingFace Inference API (free, no GPU needed).
+    Uses chat_completion (conversational) — the only supported task for Qwen2.5-7B-Instruct.
+    """
     try:
         from huggingface_hub import InferenceClient
 
@@ -169,40 +171,30 @@ def generate_with_hf_api(prompt: str, model: str = None) -> Optional[str]:
 
         client = InferenceClient(token=token)
 
-        # Use text_generation for instruct models
-        response = client.text_generation(
-            prompt=prompt,
+        # Use chat_completion — Qwen2.5-7B-Instruct is conversational only
+        # (text_generation fails: "Supported task: conversational")
+        response = client.chat_completion(
             model=hf_model,
-            max_new_tokens=1024,
+            messages=[
+                {"role": "system", "content": "You are a helpful legal assistant."},
+                {"role": "user", "content": prompt},
+            ],
+            max_tokens=1024,
             temperature=0.7,
-            do_sample=True,
         )
+        if response and response.choices:
+            text = response.choices[0].message.content
+            if text:
+                return text.strip()
 
-        if response:
-            return response.strip()
-
-        logger.warning("[HF_API] Empty response")
+        logger.warning("[HF_API] Empty response from chat_completion")
         return None
 
     except ImportError:
         logger.error("[HF_API] huggingface_hub not installed")
         return None
     except Exception as e:
-        logger.warning(f"[HF_API] Error: {e}")
-        # Try chat completion as fallback
-        try:
-            from huggingface_hub import InferenceClient
-            client = InferenceClient(token=HF_TOKEN if HF_TOKEN else None)
-            response = client.chat_completion(
-                model=model or HF_MODEL,
-                messages=[{"role": "user", "content": prompt}],
-                max_tokens=1024,
-                temperature=0.7,
-            )
-            if response and response.choices:
-                return response.choices[0].message.content.strip()
-        except Exception as e2:
-            logger.warning(f"[HF_API] Chat completion also failed: {e2}")
+        logger.warning(f"[HF_API] chat_completion error: {e}")
         return None
 
 

requirements.txt

@@ -3,6 +3,6 @@ fastapi>=0.104.0
 uvicorn>=0.24.0
 pydantic>=2.0.0
 python-multipart>=0.0.6
-huggingface_hub>=0.20.0
+huggingface_hub>=0.25.0
 requests>=2.31.0
-httpx>=0.25.0
+httpx>=0.25.0

start.sh

@@ -5,10 +5,15 @@
 # v6: Ollama is optional - HF Inference API provides fallback
 
 echo "============================================================"
-echo "  Moltbot Hybrid Engine v6.0.0 - Starting..."
+echo "  Moltbot Hybrid Engine v7.1.0 - Starting..."
 echo "============================================================"
 echo "  Timestamp: $(date '+%Y-%m-%d %H:%M:%S')"
 echo "  User: $(whoami) | Home: $HOME"
+echo "  BUILD_MARKER=2026-03-02-fix-hf-ollama-openclaw"
+echo ""
+echo "[DEBUG] Ollama check: $(command -v ollama 2>/dev/null || echo MISSING)"
+echo "[DEBUG] /usr/local/bin/ollama: $(ls -lh /usr/local/bin/ollama 2>/dev/null || echo NOT_FOUND)"
+echo "[DEBUG] Binary type: $(file /usr/local/bin/ollama 2>/dev/null || echo N/A)"
 echo ""
 
 # Optimize for HF Spaces Free tier (2 CPU, 16GB RAM)