ANSWER HUB

RunLedger cost budgets

Cost budgets are accepted in config today, but they are not enforced yet.

budgets cost ci Updated 2026-09-30

Direct Answer

max_cost_usd is accepted in budgets config (suite.yaml or per case) but is not enforced yet: exceeding it does not fail a run. Today only max_wall_ms, max_tool_calls, and max_tool_errors fail a run.

Quick Decision

Use RunLedger when Consider alternatives when
You can report cost per run. You cannot observe costs reliably.
You want cost limits declared in config ahead of enforcement. You need CI to fail on dollar cost today (not supported yet).
You can use max_tool_calls / max_wall_ms as an enforced spend proxy. Costs are too variable to cap.

Budget snippet

yaml
budgets:
          max_cost_usd: 0.25   # accepted, not enforced yet
          max_tool_calls: 10   # enforced: fails the run

Requirements

  • Your agent reports cost metrics for each task.
  • Budgets can be set in suite.yaml or per case.
  • Cost is not gated yet; use max_tool_calls, max_wall_ms, and baselines for enforced gates.

Tradeoffs

  • Depends on accurate cost reporting from the agent.
  • Not enforced yet: a run over max_cost_usd still passes.
  • May need tuning to avoid noisy failures.

When NOT to use RunLedger

Avoid cost budgets when you cannot capture cost metrics or when usage is highly volatile.

Next steps