Send prompts and images

Use Slingring when you want a second opinion, a deeper review, or a fresh ChatGPT conversation without copying context by hand.

From your terminal or local agent

slingring new-task "Review this plan and point out its biggest risks"

A local agent can run the same command. For longer prompts, ask it to write a brief to a file and send that file:

slingring new-task --prompt-file ./brief.md --reasoning-effort "Extra High"

Piped input works too:

cat brief.md | slingring new-task --stdin

Attach an image

slingring new-task "Compare these designs" \
  --image ./first.png \
  --image ./second.png

Repeat --image for multiple files. Images must be on the Mac running Slingring. Remote MCP callers cannot read or upload local files.

Get a result your agent can parse

slingring new-task --prompt-file ./brief.md --json

The JSON result includes the conversation URL, job ID, selected reasoning level, and the kind of activity observed. It does not include ChatGPT’s answer. Open the returned link to read or continue the conversation.

You can send several requests together. They queue up and start one at a time; you do not need to manage browser windows.

When a task fails

Read the error before trying again. An uncertain result means ChatGPT may have received the prompt. Check Open Latest Chat, your ChatGPT history, or slingring jobs --json first. Do not immediately create a replacement request.

The command reference covers limits and safe retries.