Instruction file imported from idangelo1/api-tester-agent-demo (.github/instructions/k6-script-generation/k6-script-generation.instructions.md). Copyright stays with the author.

k6 Script Generation Instructions

Input Confirmation

• Before generating or changing any k6 script, ask for all required parameters:
  • Base URL / environment target
  • Endpoint(s), method(s), and required headers
  • Authentication mechanism and token source
  • Load profile (vus/rps, ramp-up, duration)
  • Performance thresholds (p95, error rate, throughput)

Script Rules

• Use environment variables for URL, tokens, and credentials.
• Do not hardcode secrets in source files.
• Include checks for status code and critical response fields.
• Include thresholds for latency and error rate.
• Keep scenarios explicit and named.

Reporting Requirements

• Every k6 execution must generate report artifacts automatically.
• Always export JSON summary with --summary-export.
• Always export HTML report with K6_WEB_DASHBOARD=true and K6_WEB_DASHBOARD_EXPORT.
• Use timestamped filenames per run to avoid overwriting previous results.
• Include a request summary artifact per run (JSON or CSV) with, at minimum:
  • method
  • URL or endpoint
  • status code
  • response time
  • error message when request fails
  • response body
• Always include in the report output a dedicated section with concrete response examples from the run.
• That section must include at least 100 full response examples when the run has 100 or more requests.
• If the run has fewer than 100 requests, include all available full responses.
• Each stored example must include at minimum: method, endpoint/URL, status code, response time, response headers, and full response body.

Execution Safety

• Start with a smoke run before high load.
• Avoid destructive load against production without explicit approval.
• If required inputs are missing, stop and request them.