Skip to main content

typescript-runtime-validation

Validate a JavaScript or TypeScript project by authoring and replaying an eval plan that can build or type-check, run tests, start the project, probe endpoints, and exercise browser flows. Use this directly or when another Modernizer skill delegates its required validation.

Datos de origen

Repositorio
microsoft/upgrade-agent-plugins
Última actividad en el origen
2 de octubre de 2026 a las 22:00
Idioma detectado de SKILL.md
inglés
Estrellas
41
Forks
9

Opciones de instalación

De forma predeterminada está seleccionado el prompt que primero revisa el origen. Puedes cambiar a un comando directo o descargar una copia local.

Revisa los archivos de origen

Lee SKILL.md y los archivos complementarios que muestra SkillsMP antes de decidir si quieres instalarlo.

Explorador de archivos
9 archivos

Mostrando SKILL.md

SKILL.md
Instrucciones de origen · Vista previa de solo lectura
name
typescript-runtime-validation
description
Validate a JavaScript or TypeScript project by authoring and replaying an eval plan that can build or type-check, run tests, start the project, probe endpoints, and exercise browser flows. Use this directly or when another Modernizer skill delegates its required validation.
Validate the target project by replaying `.tsupgrader/runtime-validation/eval-plan.json` with `typescript_validate_runtime`. Here, **runtime validation means the complete project-specific validation plan**, not only checks against an already-running application. Depending on the project, the plan may install dependencies, compile or build, run tests, launch a CLI/server/app, probe routes, and replay browser flows. ## Choose the workflow - **Called from `typescript-dependencies-upgrade` or `typescript-compiler-upgrade`:** read [upgrade-workflow.md](./upgrade-workflow.md). Use `mode: "upgrade"` and preserve the caller's `rootDirectory`, `packageDirectory`, and `sessionId`. - **Invoked directly by the user:** read [standalone-workflow.md](./standalone-workflow.md). Use `mode: "standalone"` so the health check cannot create, consume, reset, or satisfy an upgrade baseline. Never use standalone mode during an upgrade, and never fabricate an upgrade workflow for a direct health check. ## Eval-plan ownership This skill owns the eval-plan authoring guidance. When the tool reports that no plan exists, follow [plan-authoring.md](./plan-authoring.md) and its linked references, then resume the selected workflow. ## Key principles 1. **Run one MCP tool at a time.** Wait for each call to complete before starting another. 2. **Keep the plan deterministic.** The same plan must be safely replayable without manual cleanup or volatile inputs. 3. **Separate plan defects from application failures.** Fix an invalid assertion rather than changing working application code to satisfy it. 4. **Do not weaken checks to manufacture a pass.** Report environment or credential blockers explicitly. 5. **Use the captured browser boundary.** For each browser flow, call `typescript_prepare_browser_recording` once and invoke the exact returned `JSTS Playwright Spec Author` child once; repeat this per flow to record multiple specs. Its final assertion creates and cleans the spec automatically; never hand-author selectors or let the child write the spec.
Ver en GitHub