# alexgreensh/token-optimizer

Find the ghost tokens. Fix them. Survive compaction. Avoid context quality decay.

Repository: https://github.com/alexgreensh/token-optimizer
Canonical: https://ross.abutalabs.com/products/token-optimizer
Language: Python
License: NOASSERTION
License Family: other
Topics: agentskills, claude-code, token-optimization, claude-code-skill, context-window, token-usage, ghost-tokens, context-engineering, token-optimizer, claude-plugin
Last push: 2026-08-26T13:43:32+00:00

## Health v2 (maintenance only)
Score: 78/100 (v2, computed 2026-09-03T02:39:23.370411+00:00)
- activity 99, release rhythm 87, longevity 13
- inputs: {"age_days": 188, "days_push": 7, "days_rel": 7, "gap_med": 0, "n_releases_24m": 238}
- flags: no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1971, forks 156 (observed 2026-08-28T04:06:00.618256+00:00)

## What it is
A Claude Code plugin/skill that finds and removes wasted 'ghost tokens' in agent context windows, helping sessions survive compaction without context quality decay. It ships variants for multiple coding agents including Claude Code, Codex, Copilot, OpenCode, and Hermes.

## Use cases
- reduce token usage in claude code sessions
- fix context quality decay after compaction
- cut llm api costs from wasted context tokens
- optimize context window for ai coding agents
- make long agent sessions survive compaction
- audit prompts for wasted tokens

## When to choose
- you use Claude Code or similar coding agents daily and hit context limits
- you want to lower token spend per session
- your agent degrades after context compaction

## When to avoid
- you don't use agentic coding tools
- you need a general-purpose prompt compression library
- you require a permissive license (license is non-standard)

## Facets
- artifact type: plugin
- maturity: active
- function: llm-inference, prompt-engineering, developer-tools, agent-framework
- domain: large-language-models, developer-tools
- platform: cli, python, cross-platform
- tags: claude-code, claude-code-skill, token-optimization, context-window, ghost-tokens, context-engineering, compaction, token-usage, ai-agents

## Member repositories
- alexgreensh/token-optimizer (main) score 78

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:06:00.618256+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T03:05:04.404911+00:00, confidence not recorded.
  - readme: https://github.com/alexgreensh/token-optimizer (fetched 2026-08-28T04:06:00.618256+00:00, sha 69d76d36862d)
- Data as of 2026-08-30T08:39:29.467469+00:00.