Most posting tools solve the wrong half of the problem. They write the text for you, then release it on a cadence that looks exactly like a cadence — same minute, same rhythm, every day.
SteamyLogic split those two jobs. Writing happened by hand, in batches, in plain .txt files dropped into a folder. A loader parsed each file, checked every post against the platform rules — 270 characters for X, 490 for Threads, no hashtags, no mentions, one emoji at most — and inserted the survivors into a SQLite queue tagged by platform.
The publishing side never wrote anything. It hashed the day's date plus the platform name into a random seed, so restarting the process on the same day returned the same slots. X got one post per day, Threads two, and the two accounts never fired in the same minute because the seed differed. Before each post it fuzzy-matched the text against the last 200 already published and dropped anything scoring 90 or higher.
What it didn't do: it only published. No reading replies, no following, no engagement metrics. Growth still depended on the operator showing up by hand.
It stopped in two steps. On 2026-05-20 Meta began rejecting the Threads token with an "Instagram blocked this action" error, and the poster exits on that rather than retrying — the last Threads post had gone out that morning. The X side kept publishing until 2026-06-08, and then the process stopped being run. The queue has sat untouched since; the posts still in it were never lost, and the date-seed and duplicate checks above are why starting it again on any later day would not repost anything.