# Skillrot

> Audits an agent skill library for context cost, dead skills and broken SKILL.md frontmatter. Use when the user asks why their agent is slow or bloated, wants to prune installed skills, asks what their skills cost in tokens, or wants a SKILL.md checked before publishing it.

- Skill: `galactic717/skillrot` (Agent Skill, multi-file: 6 files)
- Install (CLI): `npx skillmds@latest add galactic717/skillrot`
- Raw SKILL.md: https://api.skillmd.com/api/skills/galactic717/skillrot/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- License: MIT
- Author: Galactic717 (https://skillmd.com/u/galactic717)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/galactic717/skillrot

---


# skillrot

Audit a skill library: what it costs per message, what can never fire, what is broken.

## Run it

```bash
python ${CLAUDE_SKILL_DIR}/skillrot.py --json
```

Add a path to audit somewhere specific:

```bash
python ${CLAUDE_SKILL_DIR}/skillrot.py ./skills --json
```

Useful flags: `--portable` (check a skill will survive a claude.ai upload), `--no-usage`
(skip the transcript scan), `--fail-on error` (CI), `--full` (every finding).

## Reading the output

`budget.always_on_tokens` is paid on every request. `budget.body_tokens` is what loads
when skills fire and then squats in context for the rest of the session. Findings carry a
rule id, a severity and a fix.

## Reporting back

1. Lead with the always-on number and what share of the context window it takes.
2. Name the worst offenders by listing cost, marking any with `invocations: 0`.
3. List errors before warnings. Errors mean a skill is broken, not merely expensive.
4. For a prune recommendation, rank by listing cost among skills that have never fired.
   Say plainly that "never fired" is evidence from local history, not proof a skill is
   useless — a recently installed skill has not fired yet either.

Do not delete or edit any skill without the user asking for that specific change.

## Rules

`docs/RULES.md` in this directory documents every rule and cites the behaviour it catches.
Read it before explaining a finding you are unsure about.