Engineering
Overnight runs that mostly find nothing: split into scout, gate, worker, and risk
Most overnight polling runs wake up, find nothing, and still cost usage.
What hurt
le0xet (@le0xet) says 94 of 96 overnight runs find nothing but still get paid for. His fix is a custom /night-filter skill and a four-bot split: SCOUT checks for changes every 5 minutes, QUANT cancels the full run when nothing moved, EXEC does the actual work, and RISK blocks anything that can't be undone. His last night: 96 checks, 2 full runs, 0 irreversible actions. He didn't post the skill body. Not affiliated with xAI.
What to install or create
A cheap check gates the expensive run, and a separate risk role blocks irreversible steps.
Bots
SCOUT
Cheap change check.
QUANT
Cancels the full run when nothing moved.
EXEC
Does the real work only when gated in.
RISK
Blocks irreversible actions.
Plugins
None on day one. Follow the first-run steps.
Routines
None on the first run. Watch one manual pass first.
The prompt
This is a tip, not a pasteable system prompt. The author did not publish one. Follow the first-run steps.
First run
- 1Read https://x.com/le0xet/status/2110474133280547039 (credit @le0xet).
- 2Write the cheapest possible 'did anything change?' check for your overnight job.
- 3Run the full job only when that check says yes.
- 4Add a rule (or a separate risk bot) that blocks anything irreversible overnight.
- 5Compare checks vs full runs after one night and tune the check.
Approval fence
RISK blocks anything irreversible; those wait for you in the morning.
What to tweak
Steal the role split: cheap scout, gate, worker, risk blocker.
Source
le0xet @le0xet · Original post
Not affiliated with xAI. We did not write this prompt. Third-party template terms still apply.
Pro pack
Want the full setup, not just the idea?
Pro packs give you the job text, prompts, schedules and pitfalls for popular use-cases.