Managed Spend Fuse
Slops meters AI and agent usage by customer, feature, project, and model, then enforces spend policies before a noisy loop or runaway workflow becomes a surprise bill.
Spend proof is metering and control evidence. It shows how usage was counted, priced, capped, and reviewed; it does not publish private inputs, outputs, source text, private routes, or internal operating material.
Normalize provider usage into a cost ledger that can be read by customer, project, feature, and model.
Soft warnings and hard caps give operators a fuse before usage turns into uncapped spend.
Policy changes, emergency overrides, and higher-risk workflows stay inside an explicit human boundary.
Spend Fuse is offered as a managed control layer first. The base fee covers policy design and review; usage, pass-through provider cost, and exceptional operator work stay separately visible.
One provider class, two projects, up to 250k metered usage events, and weekly spend review.
Base management fee with included units, clear overage, and provider cost pass-through.
Operator-review boundary: Slops can automatically warn or block when a defined cap is crossed. Raising caps, changing billing basis, replaying disputed usage, or switching from log-ingest to proxy enforcement remains reviewed work during the managed phase.
A spend receipt is intentionally narrow. It helps answer whether the meter, control, and review trail behaved as expected without turning Slops into a prompt archive.
Receipt fields can include unit count, provider class, model class, project, feature, and price-table version.
Soft warnings, hard blocks, prepaid-credit state, and overage class are recorded as control evidence.
Public-safe proof shows metering and policy evidence only: no private inputs, outputs, retained content, private operating corpus, or internal control-plane mechanics.