Scrooge Mode — AI skill for Claude Code
Same answer, fewer tokens — KO-first LLM output-compression skill for Claude Code & Codex.
How to install Scrooge Mode
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open Kir93/scrooge-mode and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Scrooge Mode does
Same answer, fewer tokens — KO-first LLM output-compression skill for Claude Code & Codex. A Korean-native caveman alternative, measured on real session output_tokens.
Alternatives in AI
- Open Pencil — AI-native design editor 8k ★
- Paca — AI-native, free, open-source alternative to Jira, Trello, ClickUp & Monday 1.8k ★
- Squeez — Hook-based token compressor for 5 AI CLI hosts (Claude Code, Copilot CLI, OpenCode, Gemini CLI, Codex CLI) 188 ★
README
scrooge
tokens are money — spend them like a miser
Demo · Install · Surface · Benchmarks · Compared to caveman · Mechanics
Output-compression skill for AI coding agents — same answer, fewer tokens on every reply.
KO-first pentalingual (KO/EN/JA/HI/ZH) output-compression skill for AI coding agents: full on [Claude Code](https://docs.anthropic.com/en/docs/claude-code), hook + stats on Codex, skill-only on Cursor, Windsurf, Cline, Continue, Gemini CLI (see [Host support](#host-support)). The Korean register is designed around its own grammar primitives (개조식 · 음슴체 · 존댓말 제거 · 반말 default), **not** translated from English compression rules.
**`~70% KO · ~67% EN · ~65% JA · ~63% HI · ~67% ZH · single-turn chat prose · output-only`** — `claude-opus-4-8`; KO/EN N=21–24 paired median, JA/HI/ZH held-out N=10–11. **Not an agentic-session figure** — agentic work is measured separately (52% total-output savings on a 10-task corpus; the register reaches ~14–30% of billed output in real sessi
Related Skills
Lm Resizer
Rust-native context compression for Claude Code, Codex & MCP agents: filters & compresses noisy tool output (t
Tokenslim
TokenSlim — spend fewer tokens on every LLM API call: prompt compression, 350-skill router, model cascade, sem
Compressor
Reduce token usage in AI coding agents (Claude Code, Copilot, Cursor) with mode-switchable instruction packs a
Distil
Measure what context compression actually costs an LLM agent: where a session's tokens go, and whether a rewri
Claude Code Mode
Code mode for Claude Code: a bridge to Pi's codemode. The model writes one script that calls the session's too
Rustygrep
Fast grep with token-compressed, AI-native output for LLM coding agents. --llm cuts context-window tokens 60-9
Related Agents
Caveman Compress
Compression engine wrapper — runs the caveman-compress CLI on a body artifact in scratch, returns the compress
Part 19: Repowise — Codebase Intelligence For Coding Agents
60% fewer tokens. 4x faster. Your coding agents stop guessing and start knowing. ---
Token Modeller
Modelling agent that turns raw measured or inferred values into the semantic token set and emits tokens.ts, th