# Safe Change Operator

> Enforce preflight risk checks, bounded edits, and explicit verification loops for code changes. Use when tasks involve non-trivial modifications, potential regressions, or unclear blast radius.

- Skill: `arjun-bali-ds/safe-change-operator` (Agent Skill, multi-file: 4 files)
- Install (CLI): `npx skillmds@latest add arjun-bali-ds/safe-change-operator`
- Raw SKILL.md: https://api.skillmd.com/api/skills/arjun-bali-ds/safe-change-operator/raw
- Safety review: pending (external: skill-scanner PASS, skillspector PASS)
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: arjun-bali-ds (https://skillmd.com/u/arjun-bali-ds)
- Updated: 2026-09-22
- Page: https://skillmd.com/skills/arjun-bali-ds/safe-change-operator

---


# Safe Change Operator

Use this skill to prevent "YOLO" edits and silent regressions.

## Outcomes

- Assess risk before changing code.
- Keep edits bounded to explicit scope.
- Define verification criteria before implementation.
- Produce a concise safety report with rollback notes.

## Workflow

1. **Preflight risk scan**
   - Run `python skills/safe-change-operator/scripts/risk_scan.py --task "<task>" --paths <candidate files...>`.
   - Use the output score and factors as a guide, not a hard gate.

2. **Scope contract**
   - Explicitly list:
     - files to modify
     - files to read only
     - files excluded from changes
   - If score is high, ask one clarifying question before editing.

3. **Verification contract**
   - Define pass/fail checks first (tests, lint, runtime checks, snapshots).
   - Prefer reproducible commands over subjective judgments.

4. **Bounded execution**
   - Implement smallest coherent diff first.
   - Re-check risk if scope expands.

5. **Post-change safety report**
   - What changed and why.
   - Verification evidence.
   - Residual risk + rollback path.

## Output Template

```markdown
## Safety Preflight
- Task: ...
- Risk score: ...
- Primary risk factors: ...

## Scope Contract
- modify: ...
- read-only: ...
- excluded: ...

## Verification Contract
- command/check 1 ...
- command/check 2 ...

## Safety Report
- changed: ...
- verified by: ...
- residual risk: ...
- rollback: ...
```

## Notes

- For low-risk typo/doc fixes, keep process lightweight.
- For schema/auth/build/deploy edits, treat as high-risk and tighten verification.
- Read `references/risk-playbook.md` when score >= 7.