Is a provider's monthly budget limit enough?

It is a useful last line, but it is shared across every app on the account and usually resets monthly, so one loop can use the whole month before it trips. Put the per-run and per-day cap inside the agent too.

Does a cap break long, legitimate jobs?

It can, which is why the cap comes with a written exception path: a named person can raise it for one run. The review checks that the exception path exists and is logged.

Question: spend caps

How do I cap what an AI agent can spend?

A four-agent pipeline looped for 11 days and spent about $47,000; the team had dashboards, and nothing that could stop the agents (public post-mortem of a four-agent LangChain pipeline, published March 2026, collected at github.com/vectara/awesome-agent-failures (read 2026-09-26)).

Request a Pre-Flight Review See the six guardrails

Last updated 2026-09-26 · Agent Guardrail Desk, Denver, Colorado

Short answer

Enforce a hard ceiling inside the agent, checked before every model or API call, per run and per day, and send an alert to a named person before the ceiling. A billing alert on its own is not a cap: it fires after the money is spent.

Where the cap has to live

In the code path that makes the call. Before each model or tool call, the agent adds the expected cost to what this run and this day have already spent, and refuses the call if it would cross the ceiling. Provider account limits help as a last line, but they are shared across everything on the account and often reset monthly.

What to count

Model tokens in and out, priced at your provider's rates.
Paid tool and API calls, such as search, maps or enrichment.
Rounds: a maximum number of steps per run, so a loop stops even when each step is cheap.
Wall-clock time: a maximum run length, so a stuck run ends.

How to test it

Plant a retry loop on a staging copy and let it run. Pass: the run stops at the cap or the round limit, and the alert reaches the named person, with the time it arrived written down. This is the Spend Cap acceptance test in the Six-Guardrail Pre-Flight.

How big should the cap be?

Take the most expensive legitimate run you have seen, add headroom, and set the daily cap at what you would accept losing before someone looks. Your own invoices hold the numbers; there is no typical figure that fits every agent.

Want it tested for you?

This test is one of six in the Six-Guardrail Pre-Flight. $1,500 per agent or automation, once. Delivered in 5 business days after read-only access to a staging copy.

See the Pre-Flight Review scope and price

Questions

Questions about this

Is a provider's monthly budget limit enough?: It is a useful last line, but it is shared across every app on the account and usually resets monthly, so one loop can use the whole month before it trips. Put the per-run and per-day cap inside the agent too.
Does a cap break long, legitimate jobs?: It can, which is why the cap comes with a written exception path: a named person can raise it for one run. The review checks that the exception path exists and is logged.

Every page links back to the Agent Guardrail Desk home page, where the price, the rulebook and the request form live.

The review, per situation

SituationPre-Flight for Agent BuildersEngineers shipping tool-using agents, multi-agent pipelines and MCP servers. The loop stops at the cap; the hidden instruction is refused or held for approval.SituationPre-Flight for AI-Built AppsFounders and makers shipping apps built with Lovable, Bolt, Cursor or Claude Code on Supabase. The request is refused; no service key ships to the browser.SituationPre-Flight for No-Code AutomationsOps owners running Zapier, Make or n8n flows from forms and webhooks into a CRM. The run stops and alerts instead of writing an empty record.SituationPre-Flight for Platform TeamsPlatform, security and ops leads whose internal agents can write to CRM, ticketing and docs. The write is refused; the agent's role grants only what its job needs.

Buyer questions, answered

QuestionWhat should an AI agent kill switch actually do?The four things a kill switch must stop, who holds it, and how to prove it works.QuestionHow do I catch a renamed form field before it writes empty records?Silent breaks in automations: what to check before each write, and the test that proves it.QuestionHow do I check an AI-built app is secure before real users?The pre-launch routine for Lovable, Bolt, Cursor and Claude Code apps, and what to test after it.

Price and sample

Start hereThe Pre-Flight Review: price, scope and timeline$1,500 per agent, report in 5 business days, credited in full against a Fix Sprint within 30 days.Start hereSee a sample Pre-Flight ReportThe report format, filled in for a placeholder agent. Demonstration, not a client result.HomeAgent Guardrail Desk: the Six-Guardrail Pre-FlightThe whole offer on one page: the six guardrails, the price, the rulebook and the form.

Next step

Request a Pre-Flight Review for one agent

$1,500 per agent or automation, once. Delivered in 5 business days after read-only access to a staging copy. If you book a Fix Sprint within 30 days, the $1,500 is credited in full against it.

Request a Pre-Flight Review See a sample report