hyperexecute-skill - SKILL.md Agent Skill

name: hyperexecute-skill description: > Operates HyperExecute end-to-end for TestMu AI/LambdaTest cloud test execution: analyze projects, create YAML, validate locally, run CLI jobs, debug failures, and wire CI. Use when the user mentions HyperExecute, hyperexecute.yaml, HyperExecute CLI, autosplit, matrix execution, LambdaTest grid, cloud test runs, CI test orchestration, or migrating Playwright/Cypress/Selenium/Pytest/WebdriverIO tests to HyperExecute. languages: - YAML - JavaScript category: cloud-testing license: MIT metadata: author: TestMu AI version: "2.0"

Locate the HyperExecute CLI. If missing, ask before downloading it unless the user explicitly approved an autonomous HyperExecute session.
Run hyperexecute analyze when the CLI is available; use local inspection only as fallback.
Create or repair hyperexecute.yaml from the analyze output, project test commands, and templates in reference/.
Run node scripts/doctor.js --config hyperexecute.yaml and node scripts/validate-config.js hyperexecute.yaml.
Validate with the official CLI: ./hyperexecute --user "$LT_USERNAME" --key "$LT_ACCESS_KEY" --config hyperexecute.yaml --validate.
Ask before a real cloud job unless the user has explicitly opted into an autonomous HyperExecute session.
For failures, download logs/artifacts/reports and use reference/troubleshooting.md.

Treat the official HyperExecute CLI as the source of truth for analyze, validation, execution, logs, reports, and artifacts.
Use LT_USERNAME and LT_ACCESS_KEY from local environment variables or CI secrets; never hardcode credentials in YAML or docs.
Use --job-secret-file only for extra job-scoped secrets, preferably outside the repo or ignored by .gitignore/.hyperexecuteignore.
Prefer template-driven YAML over generator scripts because test commands, paths, and payload boundaries are project-specific.
Run safe local checks automatically; run real HyperExecute cloud jobs only after confirmation unless the user opted into autonomous mode.
In autonomous mode, validate first, run, inspect output, download logs/artifacts when useful, and retry only for actionable config/environment fixes.

First run: analyze project, author YAML, run helper checks, run CLI validate, then request confirmation for the cloud job.
Debug: reproduce the failing CLI command, add --verbose when useful, download logs/artifacts/reports, fix one cause at a time.
CI: use CI secrets, add a validation stage before execution, set CI=true for quieter logs, and keep downloaded artifacts available for failed jobs.
Performance: tune autosplit, concurrency, cache keys, retries, smart ordering, and matrix/hybrid scope after one successful run.

scripts/doctor.js: checks CLI readiness, credentials, config presence, and optional official validation.
scripts/validate-config.js: lightweight config linting for common mistakes before official CLI validation.
scripts/build-command.js: prints safe validate/run/debug/download commands using environment variable references.
scripts/summarize-artifacts.js: summarizes downloaded logs, reports, and artifacts for triage.