Loop/Prompt Library / Loops from builders on X
🔁 One experiment at a time, every failure written down
For a hard exact-match goal, the AI runs one experiment at a time and writes down every failed try and its lesson so it never repeats it.
Use it when: You need something to match exactly, like rebuilding old software to behave the same, and progress is slow.
🛑 Stops when: The piece passes the exact-match check, or it's parked with notes after the try limit.
Copy this loop
Goal: [what must match, e.g. 'rebuilt function gives the same output as the original']. The pass check is: [exact command or test]. First, prove the check can fail by running it on something wrong. Then work one experiment at a time. Before each try, read [EXPERIMENTS.md] so you never repeat a failed idea. After each try, add a line: what you tried, the result, and what it taught you. Save each try in version control. If you don't have enough evidence, write 'unknown' instead of guessing. Stop when the check passes, or after [10] tries on one piece; then park it with your notes and move to the next piece.
Review every change the agent makes before you keep it. Set a budget or round limit — the prompt includes one.
Found on X · by @Stenzex. Prompt rewritten in plain English by FixKit AI.
More loops like this
One useful thing every night
You give the AI a short mission, and each night it picks one task that moves you closer and leaves it for you to check in the morning.
Two watchers keep your agent healthy
Two separate AI helpers read your main agent's logs, find what broke, and must both agree on a fix before it is used.
Weekly skill cleanup
Once a week, the AI grades every saved skill or saved prompt it uses, removes the dead ones, and merges the repeats.
Overnight drafts that learn from results
Each night the AI drafts tomorrow's post from your content plan, and reads your notes about what worked before it writes the next one.
Split the work, keep only what passes
Splits a big job across several AI helpers working side by side, checks each one's work, and keeps only the work that passes.
Practice a website task until it's reliable
The AI tries a website chore, learns from each failed try, and keeps adjusting until it has steps that work every time.