Spend Control for OpenRouter
OpenRouter gives agents one API to hundreds of models — and one bill for all of them. The bill needs a budget; that's the firewall's job.
Why OpenRouter workloads overspend
Model-hopping: agents try pricier models as defaults change.
Retry loops multiply per-request costs.
A single API key can serve many agents with no per-agent budget.
How it works
sipi.bot evaluates the proposed transaction before the request hits OpenRouter: amount, merchant, category → APPROVED, BLOCKED, or FLAGGED.
Rules that fit OpenRouter workloads
Per-agent daily ceiling.
Category rule: inference budgets separate from other spend.
Velocity limit for retry loops.
Stack
Agent -> sipi.bot (decide) -> OpenRouter (route) -> model
sipi.bot caps the dollars; OpenRouter routes the requests.Compose: firewall first, router second.
FAQ
Does sipi.bot replace OpenRouter?
No — it's a decision layer in front of it. See the full comparison.
Can I budget per agent on one OpenRouter key?
Yes — per-agent rules in sipi.bot split one key's spend into budgets.
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