diff --git a/trustgraph-base/trustgraph/api/socket_client.py b/trustgraph-base/trustgraph/api/socket_client.py index 7e5dfaaf..b758f230 100644 --- a/trustgraph-base/trustgraph/api/socket_client.py +++ b/trustgraph-base/trustgraph/api/socket_client.py @@ -206,8 +206,10 @@ class SocketClient: ) else: # RAG-style chunk (or generic chunk) + # Text-completion uses "response" field, RAG uses "chunk" field + content = resp.get("response", resp.get("chunk", "")) return RAGChunk( - content=resp.get("chunk", ""), + content=content, end_of_stream=resp.get("end_of_stream", False), error=resp.get("error") ) diff --git a/trustgraph-cli/trustgraph/cli/invoke_llm.py b/trustgraph-cli/trustgraph/cli/invoke_llm.py index 261993d9..a1611625 100644 --- a/trustgraph-cli/trustgraph/cli/invoke_llm.py +++ b/trustgraph-cli/trustgraph/cli/invoke_llm.py @@ -28,7 +28,7 @@ def query(url, flow_id, system, prompt, streaming=True, token=None): if streaming: # Stream output to stdout without newline for chunk in response: - print(chunk.content, end="", flush=True) + print(chunk, end="", flush=True) # Add final newline after streaming print() else: