# Change Safety

> Plan or execute a risky scoped change where preserving user work, compatibility, verification, or rollback is the primary concern.

- Skill: `egeetas/change-safety` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds@latest add egeetas/change-safety`
- Raw SKILL.md: https://api.skillmd.com/api/skills/egeetas/change-safety/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: egeetas (https://skillmd.com/u/egeetas)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/egeetas/change-safety

---


# Change Safety

Apply this skill when modifying an existing repository, especially shared, unfamiliar, or production code.

## Before Editing

- Read applicable repository instructions and inspect working-tree status.
- Treat existing uncommitted changes as user-owned. Do not overwrite, revert, reformat, or stage unrelated work.
- Establish the current behavior from code, tests, documentation, or a reproducible command.
- Identify public APIs, persisted data, configuration, and downstream consumers affected by the change.

## During the Change

- Prefer the smallest coherent diff that solves the requested problem.
- Avoid opportunistic refactors and dependency changes unless they are necessary and explained.
- Preserve backward compatibility unless a breaking change is explicitly intended.
- For risky state changes, favor additive migration, feature flags, checkpoints, and reversible steps.
- Stop and request direction when the safe solution materially expands scope.

## Verification

Run the narrowest meaningful checks first, then broader tests in proportion to risk. Report commands, outcomes, skipped checks, residual risk, and rollback considerations.