# Targeted Validation

> Choose and run proportionate validation after a coherent code change. Use near completion of implementation, fixes, refactors, or dependency changes; not after every edit.

- Skill: `dprakash2101/targeted-validation` (Agent Skill)
- Install (CLI): `npx skillmds@latest add dprakash2101/targeted-validation`
- Raw SKILL.md: https://api.skillmd.com/api/skills/dprakash2101/targeted-validation/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: dprakash2101 (https://skillmd.com/u/dprakash2101)
- Updated: 2026-09-22
- Page: https://skillmd.com/skills/dprakash2101/targeted-validation

---


# Targeted validation

Validate after a coherent implementation and diff inspection. Choose the lowest-cost command that provides relevant confidence, then broaden only for risk, dependency reach, or failures:

```text
affected test(s) -> affected test project -> affected project build -> solution build/test
```

Use repository-provided scripts and CI-equivalent options when available. Include analyzers, formatting, contract/integration tests, coverage, or packaging only when the change or project workflow calls for them. A targeted compile during implementation is reasonable when it resolves real uncertainty.

Do not repeatedly run an unchanged expensive suite. Do not fix unrelated failures. Re-run only validation invalidated by subsequent edits.

Report exact commands and outcomes. For failures, distinguish caused-by-change, pre-existing/unrelated, and inconclusive/environmental results, with concise evidence. Never claim success for a check that was not run.