Skill detail

token-optimization

Direct Claude Code token-saving practices for recurring sessions.

MatchDirectReviewed for token optimization

Sourcehermeticormus/token-optimization-skillsExternal source

Reported installs1Popularity signal only

Inspect before use

Automated review checks relevance, not safety or endorsement. Read the source instructions before using this skill.

View on skills.sh (opens external page)Read SKILL.md on GitHub (opens external page)

Saved source preview

SKILL.md

The saved excerpt is a snapshot from review. The external source remains the complete and most current version.

---
name: token-optimization
description: Claude Code token optimization — CLAUDE_AUTOCOMPACT_PCT_OVERRIDE, MAX_THINKING_TOKENS, behavioral patterns for selective reads, grep-first discovery, and cache awareness. Use to reduce cost on multi-day projects or recurring sessions.
license: MIT
---

# Token optimization

## Two env vars

```bash
export CLAUDE_AUTOCOMPACT_PCT_OVERRIDE=70
export MAX_THINKING_TOKENS=16000
```

## Patterns

- Read with offset/limit, not whole files
- Grep before Read
- Compact at phase transitions
- Don't leave prompts cold > 5min (cache TTL)
- Separate sessions: routine vs. complex

## When to raise thinking budget

Architecture, debugging, refactoring, critical review.

## When to lower

File ops, lint fixes, trivial edits, lookups.

---

Full content at https://github.com/HermeticOrmus/token-optimization-skills.

Read the full source on GitHub

Context