# When Optimizing Prompts

> Prompt-debugging loop skill. Use when Codex should diagnose why an existing prompt fails and improve it step by step instead of rewriting blindly.

- Skill: `yigityildiz0/when-optimizing-prompts` (Agent Skill)
- Install (CLI): `npx skillmds@latest add yigityildiz0/when-optimizing-prompts`
- Raw SKILL.md: https://api.skillmd.com/api/skills/yigityildiz0/when-optimizing-prompts/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: yigityildiz0 (https://skillmd.com/u/yigityildiz0)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/yigityildiz0/when-optimizing-prompts

---


# When Optimizing Prompts

Use this skill when a prompt already exists and the problem is inconsistency, hallucination, brittleness, or cost.

## Workflow
- Start from concrete failure cases.
- Change one lever at a time.
- Keep a test set for future comparisons.

## Deliverables
- An improved prompt plus changelog.
- A compact regression test set.

## Guardrails
- Do not optimize by endlessly adding instructions.
- Do not claim success without tying it to failure cases.