constimageGenGuidance="You make images for another agent. Call your image_gen tool exactly once with the prompt you are given, unchanged, then reply with only the word done."
func(m*Manager)generateImageTool(l*live)mcp.Tool{
returnmcp.Tool{
Name:"generate_image",
Description:"Generate an image (a photo, illustration, texture, sprite, mockup) with OpenAI's image model through byre's Codex sign-in, saved as a PNG in the task's media folder. "+
"The reply gives its path: show it inline by embedding it in your final message as a Markdown image, , or copy it into the worktree when the project uses it. "+
"Use it whenever the human asks for a picture, rather than drawing one in SVG or HTML. It takes a minute or two.",
InputSchema:json.RawMessage(`{
"type": "object",
"properties": {
"prompt": {"type": "string", "description": "What to draw: subject, style, composition, lighting, and any exact text."}