bridge: add multimodal image decoding support for AGY provider

agy_bridge.py

@@ -48,9 +48,31 @@ class AGYBridgeHandler(BaseHTTPRequestHandler):
 
             system_instructions = []
             user_prompts = []
+            temp_image_files = []
+
             for msg in messages:
                 role = msg.get("role", "user")
                 content = msg.get("content", "")
+                if isinstance(content, list):
+                    for part in content:
+                        if isinstance(part, dict):
+                            p_type = part.get("type", "")
+                            if p_type == "text":
+                                txt = part.get("text", "")
+                                if role == "system":
+                                    system_instructions.append(txt)
+                                else:
+                                    user_prompts.append(txt)
+                            elif p_type == "image_url":
+                                url_data = part.get("image_url", {}).get("url", "")
+                                if url_data.startswith("data:image/"):
+                                    import tempfile, base64
+                                    header, b64_str = url_data.split(",", 1)
+                                    ext = ".png" if "png" in header else ".jpg"
+                                    with tempfile.NamedTemporaryFile(suffix=ext, delete=False) as tmp_f:
+                                        tmp_f.write(base64.b64decode(b64_str))
+                                        temp_image_files.append(tmp_f.name)
+                elif isinstance(content, str):
                     if role == "system":
                         system_instructions.append(content)
                     else:
@@ -61,11 +83,15 @@ class AGYBridgeHandler(BaseHTTPRequestHandler):
                 full_prompt += f"System Instruction:\n" + "\n".join(system_instructions) + "\n\n"
             full_prompt += "User Prompt:\n" + "\n".join(user_prompts)
 
+            if temp_image_files:
+                full_prompt += "\n\nAttached Images to analyze:\n" + "\n".join([f"- {f}" for f in temp_image_files])
+
             cmd = [AGY_BIN, "-p", full_prompt, "--output-format", "json"]
             if effort in ("low", "medium", "high"):
                 cmd.extend(["--effort", effort])
 
-            logger.info(f"Executing agy for prompt length: {len(full_prompt)} chars")
+            logger.info(f"Executing agy for prompt length: {len(full_prompt)} chars (images: {len(temp_image_files)})")
+            try:
                 proc = subprocess.run(
                     cmd,
                     capture_output=True,
@@ -73,6 +99,13 @@ class AGYBridgeHandler(BaseHTTPRequestHandler):
                     timeout=120,
                     env=os.environ.copy()
                 )
+            finally:
+                for img_f in temp_image_files:
+                    if os.path.exists(img_f):
+                        try:
+                            os.remove(img_f)
+                        except Exception:
+                            pass
 
             if proc.returncode != 0:
                 logger.error(f"agy error (code {proc.returncode}): {proc.stderr}")