Spend Control for Windsurf
Windsurf's agents run autonomously in your editor — and autonomous agents buy things. A usage meter tells you after; sipi.bot decides before.
Why Windsurf agents overspend
Agent mode runs long autonomous sessions that consume usage fast.
Agents can trigger paid tool calls and purchases beyond the editor.
Provider limits only cover one vendor; the rest is ungoverned.
How it works
Register the sipi.bot MCP tool in Windsurf. Before any spend, the agent calls sipi_guard and gets a deterministic APPROVED, BLOCKED, or FLAGGED in ~5 ms.
Rules that fit coding workflows
Per-transaction cap on compute and API credits.
Merchant allowlist for known vendors.
Approval threshold so big purchases route to a human.
Guard call
{
"tool": "sipi_guard",
"arguments": {
"amount": 1200,
"merchant": "gpu-provider.com",
"category": "compute"
}
}
# {"decision": "BLOCKED", "reason": "Merchant not on allowlist"}One tool call, one deterministic decision.
FAQ
Does sipi.bot slow Windsurf down?
No — about 5 ms per check, and only on spend actions.
Is this the same as Windsurf's usage limits?
No. Usage limits are per-vendor and reactive. sipi.bot governs transactions across every merchant before they settle.
Can it block specific merchants?
Yes — the merchant allowlist blocks anything you haven't approved.
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