A skill for Claude Code and Codex · Software and infrastructure
teste-de-carga is the skill that answers whether your site, API, or SaaS can handle N concurrent users: it converts traffic into concurrent users, RPS and a k6 rate, and picks between smoke, load, stress, spike, soak and breakpoint tests based on the question, then finds the knee, the point where latency and error rate still meet the agreed limit. At the end, the one line that sizes your infrastructure.
Solo skill: unlocks right away, no guarantee after download. Purchase terms.
Or get all 25 skills plus 9 bonuses in the Supreme Pack for R$ 497.
The problem
The question comes in concurrent users, the server feels requests per second, and the tool configures something else entirely, and mixing up those units gives you a number that looks great and is wrong. Three symptoms, and what changes with the skill.
| Without the skill | With teste-de-carga |
|---|---|
| Someone runs the test with 500 fixed virtual users and claims the system handles 500 people, not knowing a virtual user isn't a real one. | A unit contract: peak-hour users become RPS, RPS becomes an iteration rate, and only then does it calculate how many virtual users the load generator actually needs. |
| When the server chokes, the load generator itself throttles down without warning, and the report hides exactly the saturation point you were trying to find. | An open model, where requests keep arriving even when the server slows down, the way it happens in production, with dropped iterations tracked. |
| The test runs once, with the generator on the server itself, a single user, and a warm cache, and the average latency looks great. | Generator running off-target, varied data, a warm-up, three runs, p95 and p99 instead of the average, and the bottleneck cross-checked against CPU, memory, and the database. |
What you get out of it
The answer that decides your infrastructure, with the number and the reasoning behind it.
users, RPS, rate, virtual users and stages, all calculated from the peak hour
API, site, frontend under load, or breakpoint, with a lock against running against production
the highest throughput where latency and error rate still meet the limit
how many users it handles, the bottleneck, and how many instances the predicted peak requires
How it works
One sentence with the question and the latency and error limit per flow. Without a limit there's no pass criteria, just a chart.
Sessions and duration from the peak hour, not the month's average, turn into concurrent users, RPS, rate, and the stage table.
Smoke always comes first. Handling the peak calls for a load test, finding where it breaks calls for a breakpoint test after the load test passes, and degrading over time calls for a soak test.
A staging environment equivalent enough to produce real numbers, the generator running off-target with CPU headroom, third-party services excluded, and varied data.
One test type at a time, never moving on with a blown limit, the knee converted into users, and the sizing line built in with a 30% safety margin.
Before you buy
Chromium only comes into play for the hybrid test, which measures the frontend under load.
Its limits
The skill tells you how much the system handles today, where it breaks, and why. What to do about the bottleneck, whether that's cache, a bigger machine, a queue, or horizontal scaling, is the job of escala, also in the catalog.
Testing against production only happens under a well-defined set of conditions. On shared hosting without page caching, the recommendation is to turn on caching first and start at a tenth of the target, with no breakpoint test.
If the question is whether the page is fast for a single user, that's Lighthouse territory and the job of page-speed-expert, not a load test.
What you receive
The price
Just teste-de-carga
R$57one time payment
Bought it solo and now want it all? Message me on WhatsApp and I will deduct what you already paid.
Supreme Pack
R$ 1.665R$497or 10x on a card
Buying all 25 solo adds up to R$ 1.665. The pack pays for itself by the ninth skill. See everything the Supreme Pack includes
04 · Software and infrastructure
Before you ask
A concurrent user is how many people are on the site at the same time, and requests per second is what the server actually feels. One person generates several requests spread out over time, so the conversion depends on how many requests each session makes and how long it lasts. The skill does that math for you.
Only under well-defined conditions. The default path is an equivalent environment, and the scripts have a lock that refuses to run against whatever production domains you configure.
It works, carefully: turn on page caching first, start at a tenth of the target, and never run a breakpoint test, because blowing past the plan's limit turns into a 503 error and, if it repeats, can get the account suspended.
No. k6 is free and open source, and the scripts, the calculator, and the analyzer all come bundled inside the skill.
Yes, with the client's authorization for the environment being tested. The license covers your work for third parties; what's off limits is reselling or redistributing the skill.
One-time payment, 12 months of updates. No refunds after download; the Supreme Pack comes with a 7-day guarantee.
