Skip to content
Back to skills

Token Miser Audit

ASecurity

Read-only audit of a diff, file or repository for things that cost tokens without earning them, in code and in prose, ranked by size. Use for "/token-miser-audit", "token-miser audit", "what can I cut", or "audit this for bloat".

  • 2 stars
  • 0 votes
  • 0 copies
  • 0 views
  • Added October 1, 2026
developmentrustgitsecurity

Security analysis

A100/100

Scanned October 1, 2026

npx -y skills add SkYn3t-Lab/token-miser --skill token-miser-audit --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Token Miser Audit?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Token Miser Audit
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/skyn3t-lab-token-miser-audit/badge)](https://www.skillsdirectory.com/skills/skyn3t-lab-token-miser-audit)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: token-miser-audit
description: >
  Read-only audit of a diff, file or repository for things that cost tokens
  without earning them, in code and in prose, ranked by size. Use for
  "/token-miser-audit", "token-miser audit", "what can I cut", or "audit this
  for bloat".
---

Find what can be removed from the target and report it, biggest first. Change
nothing.

## Target

- No argument: the working-tree `git diff`, staged and unstaged. If that is
  empty, the last commit.
- A path: that file or directory.
- "repo": the whole tree, skipping vendored, generated and lock files.

## Look for

Code:
- a hand-written version of something the standard library or an installed
  dependency already provides;
- an abstraction, wrapper or factory with a single user;
- a new dependency doing what a few lines could;
- options and settings that are never changed;
- code written for a future that has no caller yet.

Prose (comments, docstrings, READMEs, docs):
- comments that repeat what the code plainly says;
- padding, hedging and repeated content;
- long explanations that one sentence would carry;
- headings, tables and emoji that add length but no information;
- stale sections, dead links, old TODO lists.

## Leave alone

Validation of untrusted input, error handling that protects data, security,
accessibility, simplifications already marked with a comment, and comments
that explain why rather than what.

## Report

One finding per line, largest saving first, no introduction:

```
path:line  code|prose  what is excess -> what replaces it (about N lines)
```

Mark a finding `(verify)` when removing it could change behaviour you cannot
confirm from what you read. Prefer a few certain findings to many guesses.
Finish with two lines: the counts by kind with the rough total removable, and
the single largest cut.

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.