AWS DevOps Agent and Security Agent: Read the Second Meter
Both bill by the second, both look cheap, and neither price is the whole bill. The agent-second rate covers the agent; the observability calls it makes to do its job are billed at their own rates, and the support credits that offset the first do not touch the second.
AWS ships two agents under the Frontier Agents name that are meant to act rather than suggest. DevOps Agent investigates incidents on its own, starting from a CloudWatch alarm, a PagerDuty alert, a Dynatrace problem, or a webhook. Security Agent runs penetration tests against an application you point it at and validates findings by exploiting them.
Both are metered by the second, which makes them easy to try and hard to forecast. The rates are published, so the arithmetic is worth doing before rather than after.
DevOps Agent is $0.0083 per agent-second, which is $29.88 per agent-hour. Security Agent is $50.00 per task-hour, also billed per second with partial seconds prorated. Neither has a monthly platform fee.
The rate is not the bill
The DevOps Agent FAQ says the part that matters and says it quietly: charges for connected services, CloudWatch Logs among them and trace retrievals among them, are billed separately at their own rates and are not included in DevOps Agent pricing.
That is the whole design in one line. The agent’s job is to query everything at once, so an investigation that spans four data sources is four data sources’ worth of read charges on top of the agent-seconds. If your observability bill is already the uncomfortable line on your AWS invoice, an agent that reads it aggressively is not a neutral addition.
The credits are a ceiling, not a discount
AWS gives DevOps Agent credits to customers on paid support plans, calculated from the previous month’s support charges: 100% for Unified Operations, 75% for Enterprise Support, 30% for Business Support+. The pricing page adds one sentence that changes what those numbers mean. Credits expire at the end of each month.
So this is not a discount rate applied to your usage. It is a monthly allowance whose size is set by what you already pay AWS for support, and which you lose if you do not spend it. A large Enterprise Support customer running the agent lightly is leaving credits on the table every month. A team on no paid support plan pays list, and the agent’s economics are entirely different for them than for the enterprise in the case study.
Work out your own allowance before you model anything: take last month’s support line, multiply by your percentage, divide by $29.88. That is how many agent-hours a month arrive free, and it does not carry forward.
Security Agent, priced against a thing that is hard to price
$50.00 per task-hour is the whole published rate, with the first three revalidations of any finding included and further revalidations charged by duration. The free trial is two months, up to 400 pentesting task-hours in each trial month, which is a large amount of testing to give away and a reasonable signal about the intended usage pattern: continuous, not annual.
A full run in the low tens of task-hours puts a single test in the high hundreds to low thousands of dollars. Against a scheduled human engagement, that is obviously cheaper, but I would be careful with the comparison, because the two things are not substitutes. An automated agent that maps an attack surface, works through the OWASP Top 10, and chains findings is doing the part of a penetration test that is systematic. What a good human engagement adds is the part that is not: the thing nobody thought to scope, the assumption in the business logic that only makes sense once you understand the business.
The honest framing is frequency rather than replacement. This is the tool that lets you test every release instead of twice a year. Whether it also lets you cancel the annual engagement is a question your auditor probably has an opinion about.
The scope is not limited to AWS. The FAQ says the agent can point at any application in AWS on private or public endpoints, on premises, hybrid, or in another cloud, and that authentication can come from static credentials, IAM roles, API keys, or a secret store.
Where they actually run
DevOps Agent has expanded past its launch regions. On 23 August 2026 the AWS region catalog lists it as available in eleven: N. Virginia, Oregon, Canada Central, Ireland, London, Frankfurt, Mumbai, Singapore, Tokyo, Sydney, and São Paulo.
The more interesting field is the one that is not “available.” The catalog marks Ohio, N. California, Seoul, Osaka, Stockholm, Milan, Paris, UAE, and Cape Town as not expanding, which is AWS stating that it does not plan to bring the service there rather than that it has not got round to it yet. If you are in Seoul or Ohio and were assuming this arrives eventually, that is the field to check, and it is worth re-checking, because these statuses do change.
Security Agent’s regional list is not published in its FAQ. Ask before you plan around it.
Whether to bother
DevOps Agent earns its keep if your incidents are slow in the investigation phase rather than the fix phase, and if the reason they are slow is that the data lives in four places. That is a specific shape of problem. If your incidents are slow because nobody knows who owns the service, this does not help, and neither does anything else you can buy.
If you are on Enterprise Support or Unified Operations, run it, because the credits mean you are already paying for a monthly allowance of it. Just remember that the observability reads land on a different line and no credit touches them.
Security Agent is easier to decide. Two months at 400 task-hours a month is enough to test everything you own several times over, and the answer to whether an automated pen tester finds anything in your application is not something anyone can tell you from the outside. Point it at your least-loved internal service, the one nobody has looked at since the person who wrote it left, and see what comes back.
Rates, credit percentages, trial terms, and regional statuses came from the AWS DevOps Agent and Security Agent pricing pages, their FAQs, and the AWS regional availability catalog on 23 August 2026. I have deliberately left out the preview-customer improvement percentages that circulate with these products, because they come from AWS case studies with no methodology attached and there is no way to check them.