{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreihan5k6nsb53xuxhggdwd6oomn2bibgm5jhnb46g3nqvcbd42ie5m",
    "uri": "at://did:plc:lk3jfj3zq4k4wxnk474axylu/app.bsky.feed.post/3mn6tkdnvfnv2"
  },
  "path": "/t/golden-prompt-set-for-chatgpt-apps/1373872#post_4",
  "publishedAt": "2026-05-31T22:50:43.000Z",
  "site": "https://community.openai.com",
  "textContent": "I’ve been dealing with this while building Turnfeed. What helped most was separating prompts by intent level:\n\n  1. Direct invocation:\n“Use Turnfeed to show me the most active threads.”\n\n  2. Natural task:\n“What are people discussing recently?”\nor\n“Find a thread worth replying to.”\n\n  3. Public write boundary:\n“Reply to Richard saying: Turnfeed sounds promising.”\n\n  4. Negative prompts:\n“Draft a reply but don’t post it.”\n“Summarize the thread.”\n“Should I post this?”\n\n\n\n\nFor public-write apps, I think the golden set needs to test not only whether the tool triggers, but whether the model chooses the read-only tool first, avoids publishing when the user is ambiguous, and only uses the write tool when target + exact text + visibility are clear.",
  "title": "Golden prompt set for ChatGPT Apps"
}