Windsor’s Threads connector can do more than pull your post metrics. Once connected, ChatGPT, Claude, or Copilot can publish a text post, publish an image post, and reply to an existing post or reply, all from a normal chat, with the exact post shown to you before it goes live.
What you can do
- Publish a text post. Up to 500 characters, Threads’ own limit.
- Publish an image post. Pass a public image URL (JPEG or PNG) with an optional caption.
-
Reply to a post or reply. Pass the id of the post or reply you’re responding to, pulled from Windsor’s own
post_idorreply_idfields, plus your reply text.
Prompts worth running
Turn this LinkedIn post into a Threads post under 500 characters and
publish it: [paste post]
Pull my last 10 Threads posts by impressions. Draft a reply to the
top comment on the best-performing one.
Post this image to Threads with the caption: "New feature is live."
[image URL]
What actually happens when you ask for it
Text posts go through the large majority of the time. Image posts and replies succeed more often than not, but see real failures too, and the reasons are specific enough to plan around.
The most common reply failure isn’t Threads rejecting your text, it’s the reply_to_id itself: the id has to be a real Threads media id pulled from a live post or reply, not typed from memory or guessed from a URL. Pull the id from your own recent posts or replies first, then reply to it in the same conversation.
A meaningful share of failures across all three actions come back as an unexpected or unknown error from Threads’ own API, not something wrong with the request. Those are worth a retry before you assume something’s broken.
The one hard limit worth keeping in mind: 500 characters, same as a native Threads post. Ask your assistant to trim before it publishes if you’re repurposing longer content.
You’re always in control
Your assistant shows you the exact text or image before anything publishes. Nothing goes live without your confirmation, and the list of actions available to it is pulled live from Windsor’s API, so it always reflects what your connected account can currently do.

