Stress-test evidence audit

Review existing stress-test and load-test results for resilience and availability risks.

Customize the recipe

0/1500 characters

Prompt preview

# Stress-test evidence audit

Review existing stress-test and load-test results for resilience and availability risks.

## Read-only scope
Review only the authorized app, site, server, or local source named or supplied by the user. Work from provided code, configuration, redacted exports, and completed test results. Use URLs and hostnames as scope identifiers. Request the missing artifacts if evidence is insufficient.

Use existing offline, read-only analysis on those artifacts. Exclude live probing, vulnerability reproduction, exploit payloads, credential testing, new load generation, and changes to the target or its files.

## Tools and setup · Universal
Use the linked resources as references for interpreting the supplied material. Follow the read-only scope when a linked workflow describes other actions.

- [Grafana k6 Documentation Skill](https://github.com/grafana/skills/blob/main/skills/grafana-k6/k6-docs/SKILL.md) (Skill) — Use the reporting guidance to interpret the metrics and thresholds in completed test summaries.
- [Grafana k6](https://github.com/grafana/k6) (Repository) — Review existing k6 result exports against documented metric and threshold semantics.
- [Security Threat Model](https://github.com/openai/skills/blob/main/skills/.curated/security-threat-model/SKILL.md) (Skill) — Map assets, access boundaries, and risk from supplied architecture and source evidence.

## Workflow
Analyze supplied load or stress-test summaries, thresholds, telemetry, and incident timelines. Compare response-time percentiles, error rates, recovery, and service saturation under comparable conditions. Identify where another operator-run measurement would help.

## Deliverable
A resilience audit with a metric table, breached thresholds, evidence gaps, and recommended follow-up measurements.
- Minimalism: Apply this style to report presentation only. Plain Markdown is sufficient.

## Acceptance checks
- [ ] Record the reported environment, run date, and metric units before drawing conclusions.
- [ ] Distinguish observed threshold failures and recovery behavior from hypotheses about the cause.
- [ ] Give each finding a severity, confidence, affected component, evidence reference, impact, and recommended mitigation. Explicitly list unverified controls and gaps.
- [ ] Redact credentials, tokens, personal data, and sensitive values in the report.

Recipe resources

Source, setup, and compatibility details.