- Home
- Skills
- Testing & Debugging
- k6 Load Test Writer with SLO Thresholds and Capacity Report
Works with the AI Agents you already use
k6 Load Test Writer with SLO Thresholds and Capacity Report
Find out how much traffic your app handles: realistic k6 scenarios, arrival-rate load, SLO thresholds that fail CI, first bottleneck named. READY / BOTTLENECK FOUND / NOT READY.
$5
k6 Load Test Writer with SLO Thresholds and Capacity Report
Example session with this skill installed
We're running a Black Friday sale. Last year peak was ~150 req/s on our Node API (Postgres, 2 instances on Fly). Can it take 2x? Staging is the same size as prod.
- Read your context and instructions
- Compiled the k6 load test
Load test: shop-api — VERDICT: BOTTLENECK FOUND
Question: can the API handle 300 req/s (2× last peak)? ·
Env: staging, same size as prod ·
k6: v2.3 ·
SLOs: p95 < 400 ms, errors < 1%
Scenarios
| Journey | Weight | Executor | Peak rate |
|---|---|---|---|
| browse product | 65% | ramping-arrival-rate | 195/s |
| search | 20% | ramping-arrival-rate | 60/s |
| cart | 12% | ramping-arrival-rate | 36/s |
| checkout (Stripe stubbed) | 3% | ramping-arrival-rate | 9/s |
Results
| Stage | Rate | p95 | p99 | Errors | Thresholds |
|---|---|---|---|---|---|
| smoke | 2/s | 88 ms | 120 ms | 0% | pass |
| load | 150/s | 190 ms | 410 ms | 0.1% | pass |
| stress | 210/s | 1.9 s | 6.2 s | 4.8% | fail (aborted) |
Capacity: SLOs hold up to ~190 req/s; first breach at 210 req/s.
Bottleneck
Postgres connection pool: pool size 10 per instance; at 200 req/s, /search waits 1.4 s for a connection (pool wait metric) while DB CPU is 38%. Search runs ILIKE '%term%' with no index (seq scan, 640 ms).
Fix and re-run
Added a pg_trgm GIN index and raised the pool to 25 (DB max_connections 100 allows it). Re-run:
330 req/s with p95 280 ms, errors 0.2%. Load generator CPU 41%, 0 dropped iterations.
Re-run this test against production-size data before the sale; the index migration is in migrations/20261019_search_trgm.sql.
Connects securely to your tools. The creator never sees your data.
About this skill
"Will it hold up on launch day?" deserves a number, not a guess. Most load tests do not give one: they hammer the home page with 100 virtual users, ignore think time, skip the expensive endpoints, and get bottlenecked by the laptop running the test.
This skill gives your coding agent a load-testing procedure and a reference file checked against the Grafana k6 docs, including k6 v2 changes. It:
- Defines the question and the SLOs first: target peak, safety margin, p95 latency and error-rate goals
- Models real traffic: 2 to 5 weighted user journeys with think time, test data and auth handled once per user
- Writes k6 scripts with arrival-rate executors for requests-per-second questions, per-endpoint thresholds that fail CI, checks on every response, and tagged URLs
- Runs in safe stages: smoke, load, stress/breakpoint, optional soak
- Watches the load generator so it is not the bottleneck (
dropped_iterations, CPU) - Names the first bottleneck with evidence (DB pool, slow query, CPU, downstream API, rate limiter, autoscaling lag) and re-runs to prove the fix
- Returns a capacity report and a verdict: READY, BOTTLENECK FOUND or NOT READY
Who it is for: teams preparing a launch, a sale or a marketing push, backend developers chasing slowdowns under load, and anyone who wants a performance gate in CI.
Safety: never load tests production or third-party APIs without your explicit approval of URL, rate and time window, and stubs payment, email and SMS providers.
Honesty note: it only reports numbers from runs it actually executed.
Requires explicit approval for production targets. Does not test third-party payment/email providers. k6 must be installed locally or accessible via cloud.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 28 days ago
Requires