Query tool can execute successfully at the system level, so logs show a “successful call,” but the LLM doesn’t actually receive the returned payload in a usable context unless the file is explicitly bound to the agent’s retrieval scope and referenced correctly in the prompt. Send a DM let's discuss more