/safina:load

Generate a k6 load test script tailored to what you want to test, run it locally, parse the results, and attach the report to your ticket — including a PDF summary via Chrome headless.

When to run it

Before marking a ticket QA-ready if the feature touches a high-traffic path
When performance is a stated acceptance criterion
Before a release that includes backend changes at scale
When New Relic shows degraded p95 after a deploy and you need to reproduce under load

Safety rules

Load tests are restricted to staging environments. The command enforces this — if the target domain does not contain staging, it stops.

VU limits

Virtual users	What happens
≤ 50	Default. Proceeds immediately.
51 – 250	Large. Proceeds with confirmation.
251 – 499	Warning issued. Explicit “yes” required before running.
500 – 999	Strong warning + mitigation checklist. Explicit confirmation required.
≥ 1000	Hard block. Not permitted through this command.

The default is 50 VUs — enough to surface real bottlenecks without hammering staging.

What the command does

Asks what domain, flow, and VU count to test
Validates the domain is a staging environment
Reads the existing k6 scripts in k6/scripts/ as structural patterns
Generates a new k6 script tailored to your specific test — not a health-check ping
Runs it via k6 run --summary-export=/tmp/ilmiya-load-summary.json
Parses the JSON output into a structured report
Generates a PDF via Chrome headless
Attaches the report to your active ticket (Linear or GitHub)

Script generation

The command writes a full k6 script based on your description — authentication flows, CRUD operations, multi-step user journeys, specific endpoints. The patterns come from the Ilmiya k6 library in k6/scripts/, but the script is generated fresh for your test scenario.

If auth is required, the script accepts credentials via environment variables (AUTH_TOKEN, BASE_URL) — credentials are never hardcoded in the script.

Report format

## /safina:load Report — YYYY-MM-DD HH:MM

Target: [domain]
Test: [what was tested]
VUs: [peak VUs]
Duration: [total test time]
Thresholds: [PASS / FAIL]

### Results
| Metric | Value | Threshold | Status |
|--------|-------|-----------|--------|
| Total requests | N | — | — |
| Request rate | N req/s | — | — |
| p50 latency | ms | — | — |
| p95 latency | ms | < 1000ms | pass / FAIL |
| p99 latency | ms | — | — |
| Error rate | % | < 1% | pass / FAIL |
| Check pass rate | % | — | — |

### Assessment
[What the numbers mean. Whether the service held up.]

### Recommendation
[What to do next: fix, re-test, scale, or ship.]

The PDF version of this report is attached to your ticket. If Chrome is not available, the command attaches the markdown report instead.

k6 scripts library

The k6/scripts/ directory contains reference scripts for every Ilmiya service. These are used as structural patterns when generating new tests — not run directly.

Script	Service
`accounts.js`	Accounts service
`admin.js`	Admin portal
`ai.js`	AI service
`aide.js`	Aide
`analytics.js`	Analytics
`campus.js`	Campus
`canvas.js`	Canvas
`console.js`	Console
`content.js`	Content
`edify.js`	Edify
`iam.js`	IAM
`ilm.js`	ILM core
`insights.js`	Insights
`iqra.js`	Iqra
`login.js`	Login flow
`models.js`	Models
`omni.js`	Omni
`people.js`	People
`portal.js`	Portal
`studio.js`	Studio
`users.js`	Users

Load profiles are defined in k6/common/config.js: smoke (5 VUs), default (50 VUs), large (250 VUs).

Prerequisites

k6 installed locally — brew install k6 if missing. The command checks before running.
Staging access — the test target must be a staging domain
Active ticket — the report is attached to your ticket from /safina:backlog context

Severity guide for results

Threshold	Meaning	Action
p95 > 1000ms or error rate > 1%	Service under load — not ready to ship	Fix, re-test
p95 800-1000ms	Degraded under load	Investigate before shipping
p95 < 800ms, error rate < 0.1%	Healthy	Proceed

Threshold failures block the QA-ready flag in /safina:update.