Scheduled Experiment Loops
Schedule recurring experiments with verified timing, overlap protection, auditable results, and clear stop controls.
Before & after
Same prompt, same agent — the only difference is the skill.
You could set up an experiment to test different landing page versions. You'd need to decide on your metrics, sample size, and how long to run it. You could manually check results each day or set up a scheduled report. I'd recommend using a tool like Google Analytics or Mixpanel to track your variants. You might also want to set statistical significance thresholds before you decide a winner. How would you like to structure the experiment?
AUTONOMOUS EXPERIMENT LOOP CONFIG
SETUP
• Interval: Daily (UTC 09:00 check-in)
• Duration: 14 days or statistical significance (p < 0.05), whichever comes first
• Variants: 3 landing pages ("Simple headline", "Social proof lead", "Free trial CTA")
DISTRIBUTION
• Even split: 33% traffic each
• Min sample size per day: 200 views per variant
METRICS TRACKED
• Primary: Conversion to signup (target: variant A 8%, baseline 6%)
• Secondary: Form completion rate, time-on-page
DAILY REPORT
Timestamp, variant name, conversions today, cumulative rate, confidence interval, "on track?" (yes/no)
Trigger escalation if: Variant underperforms baseline by > 2% on day 7
WINNER CRITERIA
• Primary metric reaches 95% confidence (Bayesian)
• No red flags on secondary metrics
• Deploy winner to 100% traffic automatically
HOOK: CronCreate schedules daily check; result emailed to teamAbout this skill
Your experiment works once, but making it run every ten minutes introduces new problems: overlapping runs, duplicate results, unclear stopping behavior, and schedules that disappear when the session ends. This skill turns a defined experiment into a recurring job with explicit timing, bounded execution, and a verifiable record of what happened.
What it does
- Converts a requested interval into a supported schedule, checking timezone and calendar behavior where relevant.
- Uses CronCreate when available and checks its actual scheduling and persistence capabilities before promising unattended execution.
- Detects existing equivalent jobs before creating another copy.
- Builds run instructions that prevent overlap, identify duplicate invocations, preserve results, and respect execution limits.
- Reads back the saved job and distinguishes registration from successful experiment execution.
- Provides the job identifier, result location, and supported way to stop the loop.
What you get
A configured recurring experiment when scheduling tools are available, plus a concise receipt showing its timing, execution limits, evidence, and cancellation controls. If the environment cannot support the requested loop, you get a clearly labeled unscheduled specification and the precise capability or decision needed to activate it.
Best for
Developers and researchers who already have a runnable experiment and want to repeat it every few minutes, hourly, daily, weekly, or monthly without losing control of execution or results.
SKILL.md preview
---
name: scheduled-experiment-loops
description: Schedule a defined experiment with CronCreate when available, verify its timing and lifetime, and add overlap protection, run records, execution limits, and cancellation controls. Use when the user says "run this experiment every ten minutes", "keep this research cycle running hourly", "make this experiment run daily", "schedule weekly experiment runs", "repeat this monthly", "my experiment loop keeps running twice", "the scheduled experiment stopped overnight", or "/ar:loop".
---
# Scheduled Experiment Loops
… (sign up to view the full skill)More development skills
View all Development skills →Performance Bottleneck Profiling: Node, Python, Go
Classifies a slowdown, gives exact profiler commands, reads the output, and proves the fix with same-conditions p95/p99 numbers.
Feature Scaffolding with Working Tests
Turn a short feature spec into code, runnable tests, and integration wiring that match your existing project.
Feature Flag Rollouts and Kill Switches
Design staged flag rollouts, verify rollback paths, and audit flag debt with explicit defaults, gates, and retirement plans.
Microcontroller Firmware and RTOS Engineering
Build and debug MCU firmware with bounded ISR/RTOS timing, safe DMA ownership, and measured power budgets.