context-budget-audit

What problem does it solve?

Context window capacity is limited and many Claude Code components (agents, skills, MCP servers, hooks, CLAUDE.md) cumulatively consume tokens that degrade performance, increase costs, and shorten session usefulness. This Skill estimates token consumption per component, detects bloated files and configurations, and surfaces high-impact opportunities to reclaim context budget.

Core Features & Use Cases

• Per-component token estimation using prose and code heuristics to convert file size and structure into token estimates.
• Aggregated budget reporting that ranks agents, skills, MCP schemas, CLAUDE.md, hooks, memory, and docs by token cost and savings potential.
• Prioritized recommendations with actionable mitigations such as trimming SKILL.md preambles, enabling progressive loading, and consolidating agents.
• Use Case: Run after adding new agents or MCP tools to quickly find which additions pushed your total into a warning or red threshold and get a prioritized remediation plan.

Quick Start

Run a full context budget audit across agents, skills, MCP servers, hooks, and CLAUDE.md to produce a token usage summary and prioritized savings recommendations.