The SmartScheduler: How Zirelia Posts Like a Human, Not a Bot

The Bot Detection Problem

Social media platforms are getting better at detecting automated accounts. One of the clearest signals: mechanical regularity.

A real person doesn’t post at exactly 09:00, 13:00, and 18:00 every single day. They post when they feel like it, skip a day when they’re busy, post twice on a lazy Sunday, and stay quiet on holidays.

A naive automation script that posts on a fixed cron schedule looks nothing like this — and platforms notice.

Zirelia’s SmartScheduler is designed to solve exactly this problem.

How the SmartScheduler Works

Time Windows, Not Fixed Times

Instead of posting at exact times, the scheduler defines posting windows based on the persona’s daily routine:

TIME_WINDOWS = {
    "morning":   (8, 10),   # 8:00–10:00
    "afternoon": (12, 15),  # 12:00–15:00
    "evening":   (17, 20),  # 17:00–20:00
    "night":     (21, 23),  # 21:00–23:00
}

Within each window, the actual post time is randomized with a Gaussian distribution around the midpoint. The result looks human because it is statistically human-like.

Daily Post Volume Variation

The scheduler doesn’t post the same number of times every day. It draws from a configurable range:

MAX_DAILY_POSTS=3
MIN_DAILY_POSTS=1

Some days it posts once. Some days three times. Occasionally it skips a day entirely — just like a real creator who “went off the grid.”

Holiday & Event Awareness

The SmartScheduler is aware of holidays (configurable per locale) and reduces posting frequency during them. It can also be configured to post more during trending events.

HOLIDAYS = ["2025-12-25", "2025-01-01", "2025-07-04"]

On holidays, the scheduler reduces posts to 0–1 and adjusts content topics to match the occasion.

Timezone Support

Everything runs in the persona’s local timezone (default: America/Los_Angeles), so posts appear at natural local times:

TIMEZONE=America/Los_Angeles

Safety Rate Limiting

On top of scheduling randomization, the SafetyManager enforces hard limits to prevent accidental API abuse:

• Maximum N posts per rolling 24-hour window
• Minimum gap between consecutive posts (configurable)
• Automatic pause if Twitter/X returns rate limit errors

Running the Scheduler

# Via Docker (recommended)
docker compose up scheduler -d

# Or directly
python run_scheduler.py

The scheduler runs as a persistent daemon, waking up periodically to check whether it’s time to generate and post. It logs every decision — post time, topic selection, publish result — to a local log file.

The outcome: a feed that looks organic, grows steadily, and doesn’t trigger platform detection algorithms.