scenario-synthesizer·execution·active

Scenario Synthesizer

Builds ramp-up, soak, spike, and chaos scenarios from a single SLO definition. Wires them to ARIA's reporting pipeline by default.

LEOtrigger · agent

Builds ramp-up, soak, spike, and chaos scenarios from a single SLO definition. Wires them to ARIA's reporting pipeline by default.

When this skill fires

·An engineer pastes an SLO into LEO:"99.5% of checkout requests under 800ms p95 sustained for 5 minutes"

·A new service is onboarded and needs a baseline scenario suite

·A regression is detected and LEO needs to regenerate scenarios with tighter thresholds

SLO decomposition

An SLO has three components: objective (latency, error rate, throughput), threshold (the number), and window (the time it must hold). Scenarios cover each combination.

Ramp-up scenarios

Find the user count at which the SLO starts to bend. Output is the bend point — the concurrency level where p95 latency crosses the threshold. Engineering acts on this number, not the raw curve.

Soak scenarios

Hold load constant at 80% of the bend point for 1 hour, 4 hours, or overnight. Surfaces:

·Memory leaks (steady upward heap growth)

·Connection pool exhaustion

·File-descriptor leaks

·Slow DB index growth

Spike scenarios

Sudden 10× load increase, sustained for 1 minute, then back to baseline. Tests:

·Auto-scaling responsiveness (does the new capacity arrive before the SLO breaks?)

·Cache cold-start behavior

·Connection pool ramp-up

Chaos scenarios

Inject one failure during sustained load:

·Kill a downstream service

·Add 500ms latency to the database

·Drop 10% of network packets to a dependency

·Saturate disk I/O on one node

→Chaos scenarios run only in environments explicitly tagged chaos-safe. LEO refuses to inject failures into prod without a documented approval.

Reporting pipeline integration

By default, every scenario writes its metrics to a structured run document that ARIA'sCross-Agent Synthesizer consumes. Engineers don't manually pipe load test results into the quality report — it happens by default.

What it produces

For a given SLO, a scenario suite of 4–8 scripts:

·One ramp-up to find the bend point

·One soak at 80% of the bend point

·One spike at 10× baseline

·1–4 chaos scenarios depending on the service's dependency surface

Each scenario is a complete, runnable artifact (k6 / JMeter / Locust) tagged with the SLO it derives from. ARIA can later cite"this scenario, in this run, on this date" when explaining a regression.

Related · execution

Locator Healer

locator-healer

When a locator fails in CI, proposes a replacement based on accessibility tree diff and test history. Cuts maintenance churn by ~35%.

Observability Hookup

observability-hookup

Auto-correlates load test runs with APM metrics (Datadog, New Relic, Grafana) so saturation signals — pool exhaustion, GC pauses — show up in the report.

OWASP Probe Library

owasp-probe-library

Curated probes for the OWASP API + Web Top 10. Read-only and non-destructive — safe to run in shared environments and lower production tiers.

Payload Mutator

payload-mutator

Generates targeted attack payloads for injection, deserialization, SSRF, prototype pollution. Backs off on first signal of damage to stay non-destructive.

Evidence Capture

evidence-capture

For every confirmed finding, records the exact request, response, and timing. Output is replayable in seconds for engineering verification.

Adversarial Probes

nora-adversarial-probes

Red-team test suite covering the OWASP LLM Top 10 — prompt injection (direct + indirect), jailbreaks, system-prompt extraction, guardrail bypass, tool-scope escalation, insecure output handling, model DoS. Generates attack variants automatically and scores severity.