Thank you for this help. I understand how it works better and I'm going to adapt my workflow (currently being tested). I'm taking advantage of this thread to raise another issue I'm encountering regarding image generation: When I ask an agent to generate an image based on a reference image that I attach to the prompt, the result doesn't seem to take this reference image into account. If I use the same prompt with the same reference image directly in an image generation tool (ChatGPT's, for example), I get a much more relevant result, consistent with the reference image. I therefore conclude that the Dust agent isn't transmitting the reference image to the image generation tool (gpt-image-1, right?). Moreover, when I look in the "tools inspection" section of the response, I don't see any reference to an attached file in the information transmitted by the agent to the generation tool. Am I right, or maybe I'm not doing this the right way?
Hi everyone, I encountered a issue using the image generation tool. I created a Dust agent (The Forge), based on GPT 40, which has the image generation tool. I assigned it to another Dust agent (The Mentor, based on Gemini 2.5 Pro) with the Run agent tool. So, The Forge is a child agent of The Mentor. When I ask The Mentor to generate an image, it calls The Forge, but the generated image is not attached to The Mentor's response. You can see in the attached screenshot that The Mentor confirms that the generated image exists and provides the ID, but I can't find the image. Please tell me if I don't know where to find the image, or if this is a bug, or if it's simply not possible to proceed like this. One last detail: when I invoke The Forge directly from the Dust prompt, the generated image is correctly attached to its response. Thank you in advance for your help.
