Audit Claude Code token consumption across context sources and MCP servers.
npx skills add https://github.com/haingt-dev/agent --skill token-optimize
Please help me install this Agent Skill. Skill: token-optimize Source: https://github.com/haingt-dev/agent/tree/main/global/skills/token-optimize Command: npx skills add https://github.com/haingt-dev/agent --skill token-optimize
Audit Claude Code token consumption across project and global context. Measures all context sources, benchmarks against official limits, detects duplication and waste patterns, produces actionable recommendations with quantified savings. Use this skill whenever the user mentions token costs, context size, wants to optimize their Claude Code setup, or complains about expensive sessions. Triggers: "token optimize", "audit tokens", "context audit", "optimize context", "reduce tokens", "token usage".
Run a token audit on the current workspace to generate a token-audit.json snapshot.
High-intent search queries and answers about installing and using this skill.
Token optimization in Claude Code analyzes all always-on context sources including CLAUDE.md, memory, and MCP servers to identify token waste and generate prioritized recommendations for reducing context size without losing meaning.
A Claude Code token audit analyzes all always-on context sources including CLAUDE.md, memory files, skill descriptions, and MCP server definitions to quantify duplication and waste patterns across baseline versus runtime.
To reduce token costs and optimize context size, this skill benchmarks context against official limits, detects duplication, and outputs a prioritized plan with estimated savings and a token-audit.json snapshot artifact.
A Claude Code context audit outputs a token-audit.json snapshot artifact alongside a contextual breakdown and prioritized recommendations quantifying token savings across baseline versus runtime.
Expensive Claude Code sessions are often caused by duplicated or bloated context in CLAUDE.md, memory, and MCP definitions; this skill audits those sources to detect waste patterns and quantify actionable token savings.
Yes, this context audit works with MCP servers and memory-bank files by measuring their token consumption alongside CLAUDE.md and skill descriptions to benchmark usage and detect duplication against official limits.
Alexander-TyagunovInitialize Magician workspaces with directory structure, CLAUDE.md, and permissions allowlist.
kmshihab7878Estimate token consumption across Claude Code components and output prioritized savings recommendations.
pur3v4d3rConfigure MCP servers in .mcp.json or plugin.json to expose external tools.
georgekhananaevCompress Markdown and text into token-efficient formats with Python scripts.
