Audits and optimizes token usage across Claude Code components to improve performance and reclaim context space.
Context Budget is a specialized diagnostic tool for Claude Code that identifies token bloat across agents, skills, MCP servers, and project rules. By performing a multi-phase inventory and classification of every loaded component, it surfaces actionable recommendations to reduce overhead, prevent performance degradation, and maximize the available context window for actual coding tasks. It is essential for developers managing complex setups with numerous tools and custom instructions who want to ensure their AI assistant remains fast, responsive, and accurate.
主要功能
01Prioritized optimization reports with estimated token savings
02MCP server overhead estimation and CLI-replacement suggestions
03172,010 GitHub stars
04Automated token consumption auditing for agents, skills, and rules
05Context headroom analysis for planning future tool expansions
06Detection of redundant components and bloated frontmatter descriptions
使用场景
01Troubleshooting sluggish Claude Code sessions or declining output quality
02Auditing context overhead after adding multiple new MCP servers or skills
03Planning complex project setups to stay within the model's token window
