Optimizes Claude Code performance by auditing and reducing token consumption across agents, skills, and MCP servers.
The Context Budget skill is a performance optimization tool designed to maintain high-quality AI outputs by managing Claude Code's context window. It scans all loaded components—including agents, skills, MCP servers, and rules—to identify bloat, redundancy, and unnecessary token overhead. By providing a detailed report and prioritized optimization suggestions, it helps developers reclaim context space, ensuring that the model remains fast and focused even in complex, multi-agent environments.
主要功能
01Detailed MCP server auditing to identify high-overhead tool schemas
02Automated token estimation for agents, skills, and rule files
03Prioritized optimization reports with estimated token savings
04172,008 GitHub stars
05Detection of redundant components and bloated agent descriptions
06Verbose analysis mode for file-level breakdowns and logic overlaps
使用场景
01Planning context headroom before scaling up a project with new automated workflows
02Improving AI response quality and speed when the session feels sluggish
03Auditing the impact of adding new MCP servers or multiple specialized agents
