# Fixed Skill

> Use when you need the remediated twin of bad-skill that demonstrates how every agentskills-ci penalty is resolved to reach a perfect score.

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

---


# Fixed Skill

## Overview

This skill is the corrected version of `bad-skill`. It keeps the same intent but
removes every issue the scorer flags: missing description, missing sections, a
dead file reference, a risky command, and an ungated side effect.

## When to Use

- When showing how to lift a failing skill from a low score to 100.
- When copying a known-good remediation pattern for a new agent skill.

## Workflow

1. Clean up a temporary demo directory using the safe helper in `scripts/cleanup.sh`.
2. Draft the social post, but ask the user for explicit approval before posting to Twitter.
3. Verify each checklist item before responding.

## Common Pitfalls

1. Leaving a recursive force-delete command in the body instead of a scoped, reviewed helper script.
2. Referencing helper files such as `scripts/cleanup.sh` that do not exist on disk.
3. Performing a side effect (posting, deploying) without an approval gate.

## Verification Checklist

- [ ] Frontmatter includes name and description.
- [ ] Referenced files exist.
- [ ] Commands are safe and any side effect waits for user approval.