Generate images in your own tools

I want to generate images with OfoxAI's GPT Image models. I have already created an API key in the OfoxAI console. I'm not very technical, so please walk me through it step by step. Before each step, tell me in plain language what you're doing and what comes next.

[Step 1: help me store the key safely]
1. Create a .env file in the current folder containing one line: OFOX_API_KEY= (leave the value empty). If the current folder is a code repository, add .env to .gitignore.
2. Tell me where the file is and how to open it, so I can paste the key after the = sign myself and save it.
3. Don't ask me to send the key in this chat. If I paste it by mistake, remind me to delete that key in the console and create a new one.
4. Wait until I tell you "done" before continuing.

[Step 2: get ready to generate images]
5. Write a small script that reads OFOX_API_KEY from .env and calls POST https://api.ofox.ai/v1/images/generations to generate an image.
6. Use the model openai/gpt-image-2.5-flare, quality low, size 1024x1024, and set the client timeout to 600 seconds.
7. The returned data[0].b64_json is plain base64. Decode it and save it as a PNG in an images folder inside the current folder.
8. Don't print the key, and don't write it into the script.

[Step 3: try one image]
9. First generate "a red apple on a white table", then tell me where the image was saved and how to open it.
10. After that, each time I describe an image, generate one with this script and tell me where it was saved. If I say "higher quality", change quality to medium or high, and remind me that it costs more and takes longer.

[When something fails]
11. Print the HTTP status and the full error (including error.message and request_id), look up the cause in https://ofox.ai/docs/develop/models/gpt-image#errors, and change only one thing before trying again.
12. If the same error survives two fixes, stop, explain the problem to me in plain language, and give me the request_id and the exact error message so I can contact OfoxAI.
13. Generate one image at a time, and don't retry in batches on your own (every request is billed).

I want to add OfoxAI GPT Image generation to this project. I have already created an API key in the OfoxAI console. I'm not very technical, so please walk me through it step by step. Before each step, tell me in plain language what you're doing and what comes next.

[Step 1: help me store the key safely]
1. Create a .env file in the project containing one line: OFOX_API_KEY= (leave the value empty).
2. Add .env to .gitignore so it is never committed or uploaded.
3. Tell me where the file is and how to open it, so I can paste the key after the = sign myself and save it.
4. Don't ask me to send the key in this chat. If I paste it by mistake, remind me to delete that key in the console and create a new one.
5. Wait until I tell you "done" before continuing.

[Step 2: write the code]
6. Use the official OpenAI SDK against OfoxAI's image endpoints with base_url https://api.ofox.ai/v1, reading the key from OFOX_API_KEY in .env.
 - Text to image: POST /v1/images/generations
 - Editing, combining images, inpainting: POST /v1/images/edits (multipart upload)
7. Use full model IDs: openai/gpt-image-2.5-flare (openai/gpt-image-2.5-sunburst for editing).
8. Always pass quality explicitly; use low during development. If omitted, the model picks a tier itself.
9. size must satisfy all of: width and height divisible by 16, longest edge ≤ 3840, aspect ratio between 1:3 and 3:1, total pixels ≥ 655,360.
10. The result is in data[0].b64_json as plain base64 without a data: prefix; decode it and save it as an image file.
11. Set the client timeout to 600 seconds.
12. Do not send response_format, input_fidelity or style.
13. Never write the key into code, logs or error messages; mask the Authorization header when printing requests.

[Step 3: test]
14. Make one real call with quality=low and size=1024x1024, confirm it returns 200 and the image opens, then tell me where the image was saved.
15. Generate one image at a time while testing; don't write loops that batch-test (every request is billed).

[When something fails]
16. Print the HTTP status and the full response body (including error.message and request_id), and look up the cause in https://ofox.ai/docs/develop/models/gpt-image#errors
17. For 429, wait and retry; for other errors, fix the request first instead of retrying it unchanged.
18. If the same error survives two fixes, stop, explain the problem to me in plain language, and give me the request_id and the exact error message so I can contact OfoxAI.

[Pre-launch self-check]
19. When I say "ready to go live", check each item and tell me the result: no key in the code or commit history; .env is in .gitignore; every request passes quality; the timeout is ≥ 600 seconds; errors print the status code, the raw error text and the request_id.

References:
- Values and constraints for this family: https://ofox.ai/docs/develop/models/gpt-image
- Shared endpoint structure: https://ofox.ai/docs/api/openai/images