Is sipi.bot Worth It?
Worth it if your agents can spend money — and they can. $99/mo is one documented runaway-agent incident's rounding error. Here's the honest math.
The cost side
Team is $99/month with unlimited evaluations. Self-hosting the MIT core is free.
The risk side
The public incident database documents real runaway events — from hundreds to millions of dollars. The founder story is a $12,400 single night.
If your agents hold payment methods, the question isn't whether spend control pays for itself — it's how quickly.
The honest caveat
If your agents can't spend money at all (no payment rails, no paid tools), you may not need it yet. The moment they can, you do.
Evaluate free first: self-host the core or run the playground, then decide.
At a glance
| Scenario | Without firewall | With sipi.bot |
|---|---|---|
| Agent with payment rail | Incident risk | $99/mo, unlimited evals |
| Agents with paid tools | Unreviewed spend | Caps + approvals |
| No spend-capable agents | No exposure | Not needed yet |
FAQ
What does $99/mo buy exactly?
Unlimited evaluations, the dashboard, managed approval queue, and stored audit log — plus the entire deterministic decision engine.
Can I try it before paying?
Yes — the playground runs live checks free, and the MIT core self-hosts free.
What's the cheapest real protection?
Self-host the free core and enforce two rules: a per-transaction cap and a velocity limit. That covers the two most common runaway patterns.
Related
Stop the next $12,400 night.
One API call (or MCP tool) in front of every agent transaction — APPROVED, BLOCKED, or FLAGGED, deterministic, ~5 ms, fully logged.
See plans — from $99/mo Try a live check