Draft image names and categories from visible content
Provide authorized test images, filenames and naming rules. Use an image-capable chat to suggest names and categories with visible evidence.
Execution environment and usage
AI / software
ChatGPT file/image inputs (optional draft-analysis entry)
Where to operate
ChatGPT new conversation, + → Add photos & files for supplied documents/images.
Getting started
- Start a new ChatGPT conversation, choose + → Add photos & files, and select the required files.
- Provide authorized test images plus a separate image-ID/original-filename mapping and the original prompt’s scope or naming requirements.
- Confirm readable images/fields, then check source links, unknowns and mappings in the draft. Do not send messages or modify originals.
Setup and plan requirements
Authorized images, original filename list, naming rules, and sufficiently readable resolution.
Prompt portability limits
Image understanding is required; agent browser/control is not. Text-only chat can only use user-supplied descriptions, not inspect unseen images.
Usage limitations
Retirement notice in current official help; retain historical attribution, check Work separately for any new implementation. Historical tool/source attribution is retained. Not hands-on tested; verify plan, region, workspace and site access separately. Image filenames/metadata are not reliably visible to the model; provide an explicit separate mapping.
Environment sources
- https://help.openai.com/en/articles/11752874-chatgpt-agent · 2026-10-19
- https://help.openai.com/en/articles/20001275-chatgpt-work-and-codex · 2026-10-19
- https://help.openai.com/en/articles/20001280-using-cloud-browser-in-chatgpt · 2026-10-19
- https://help.openai.com/en/articles/8555545-uploading-files-and-audio-to-chatgpt · 2026-10-19
- https://help.openai.com/en/articles/8400551-chatgpt-image-inputs-faq · 2026-10-19
適用情境 / When to use
Provide authorized test images, filenames and naming rules. Use an image-capable chat to suggest names and categories with visible evidence. Do not identify people or change originals; review the list before applying it. Source record: This commenter’s personal account supplies no independent accuracy measurement.
步驟 / Steps
- Prepare licensed nonsensitive test images, a filename manifest and an organizing purpose.
- Separate observable objects, scenes and uncertain descriptions; retain original-name mappings.
- Check collisions, omissions and description errors before applying changes.
提示詞 / Prompt
Using [licensed test images/original filename manifest/organizing purpose/naming rules], propose an image catalog. Describe only visible content, mark uncertainty and retain original names, proposed names and classification reasons. Check collisions and omissions. Do not identify people, infer sensitive traits or alter originals.
留意事項 / Cautions
- Do not identify people or infer sensitive traits.
- Do not move originals or upload private albums.
來源限制 / Evidence limits
Descriptions may misidentify objects; successful sorting establishes neither identity nor completeness.
流程說明 / Workflow note
These steps and prompts are original, untested exercises by this site, not the author’s verbatim workflow. Reported results do not guarantee exercise outcomes.
原始工具組合 / Original tool stack
ChatGPT Agent; image archive
Source checked: 2026-10-20