Scale computer-use 2.0 with open-source drivers, cross-OS sandboxes, and benchmarks for training, evaluation, and data generation.
Rust agentai-agentapplecomputer-usecomputer-use-agentcontainerizationcuadesktop-automationhacktoberfestlumemacosmanusoperatorswiftvirtualizationvirtualization-frameworkwindowswindows-sandbox
Fix multimodal user inputs in the anthropic loop
Dillon DuPont2dfaf0047da6
diff --git a/libs/python/agent/agent/loops/anthropic.py b/libs/python/agent/agent/loops/anthropic.pyindex 306c5f334d..cd616d385f 100644--- a/libs/python/agent/agent/loops/anthropic.py+++ b/libs/python/agent/agent/loops/anthropic.py@@ -132,23 +132,22 @@ def _convert_responses_items_to_completion_messages(messages: Messages) -> List[converted_content = []for item in content:if isinstance(item, dict) and item.get("type") == "input_image":- # Convert input_image to Anthropic image format+ # Convert input_image to OpenAI image formatimage_url = item.get("image_url", "")if image_url and image_url != "[omitted]":- # Extract base64 data from data URL- if "," in image_url:- base64_data = image_url.split(",")[-1]- else:- base64_data = image_url-converted_content.append({- "type": "image",- "source": {- "type": "base64",- "media_type": "image/png",- "data": base64_data+ "type": "image_url",+ "image_url": {+ "url": image_url}})+ elif isinstance(item, dict) and item.get("type") == "input_text":+ # Convert input_text to OpenAI text format+ text = item.get("text", "")+ converted_content.append({+ "type": "text",+ "text": text+ })else:# Keep other content types as-isconverted_content.append(item)