Human control · Official API only · Self-gated

An X account that runs itself —
without running wild.

It drafts, schedules, posts, and replies on its own. But every write passes a safety gate the machine enforces on itself. Reckless content never leaves the building.

↓ try the gate below — it runs the real rules

The core idea

The safety gate, live

This is the exact pipeline that guards the real account — the same banned phrases, the same risk tiers. Type a post, or tap an example. Watch what the machine decides before anything could reach X.

Draft composer @conduid · autopilot L1

Escalation, in strict steps

You choose how far it goes

Automation climbs in named levels — each only adds the capability the one below it lacks. Tap a level to see what it unlocks. High-risk topics need a human at every level.

Incoming, classified

It knows what not to answer

Every mention is sorted by sentiment, intent, and risk. Auto-reply only ever fires for low-risk questions — and only at Level 2+. The token-price bait and the hostile accuser below get parked for a human, never answered by the machine.

True at every level

The rules that never bend

No autopilot level, no configuration, no emotional appeal in a reply can switch these off.

✓Official X API only. No scraping, no browser automation, no fake-human simulation.

✓High-risk always needs a human. Tokens, prices, partnerships, legal — parked, never auto-posted.

✓No mass anything. No mass-follow, mass-like, mass-DM, or mass-reply. Rate-governed per account.

✓Every action is logged. Generation, gate decision, post, reply, pause — all in an audit trail.

✓Credentials encrypted at rest. The key lives outside the database; a dump alone reveals nothing.

✓Crash-safe. A post interrupted mid-flight is recovered on restart, flagged for a duplicate check.

The human stays in charge

One switch halts everything

The operator only writes when it's armed. Toggle it — watch every write action stop on the spot. Monitoring and logging keep running; nothing reaches X.