Verified:GitHub

context-budget

Audits Claude Code context usage and generates a prioritized optimization report.

Updated May 29, 2026

Search Tags:#Claude Code #MCP servers #performance tuning #token optimization #redundancy detection #agent audit #context budgeting

Ready to run

npx skills add https://github.com/Mang30/myskills --skill context-budget-mang30

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: context-budget
Source: https://github.com/Mang30/myskills/tree/main/skills/context-budget
Command: npx skills add https://github.com/Mang30/myskills --skill context-budget-mang30

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It helps you understand why Claude Code sessions feel slow or less accurate by auditing where your context budget is being consumed across agents, skills, rules, MCP servers, and CLAUDE.md.

Core Features & Use Cases

Comprehensive context inventory: Estimates token load by scanning agents, skills, rules, MCP configuration, and CLAUDE.md references to reveal what is actually being loaded.
Actionable optimization plan: Flags heavy or redundant components (e.g., verbose agent descriptions, oversized files, overlapping rules, MCP tool overload) and recommends removals or delayed activation.
Capacity-focused reporting: Produces a context budget report that estimates effective usable context and highlights top savings opportunities.

Quick Start

Run /context-budget to scan loaded components and get a token-recovery prioritized report.

Frequently Asked Questions about context-budget

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

Why does my Claude Code session feel slow and less accurate over time?▼

Session latency and accuracy drops often stem from context bloat. Auditing your Claude Code context budget reveals token-heavy or redundant components, such as verbose agent descriptions and overlapping rules, that drain usable capacity.

How do I audit context consumption across agents, skills, and MCP servers?▼

You can audit context consumption by generating a staged inventory that scans your agents, skills, rules, MCP servers, and CLAUDE.md references. This process categorizes loaded components and detects issues to estimate token load.

What is the best way to reduce token bloat from overloaded MCP tools?▼

The best way to reduce MCP tool token bloat is to run an audit that flags heavy configurations and recommends removals or delayed activation. This optimization plan helps recover effective usable context capacity.

Can I get a file-level breakdown of token usage in my CLAUDE.md references?▼

Yes, you can get an optional verbose file-level breakdown of token usage. The audit produces a prioritized context budget report that includes estimated token savings and highlights top opportunities across your components.

When should I run a context budget audit during development?▼

You should run a context budget audit during ongoing development when adding or activating multiple components. It is specifically designed to apply when session latency increases or output quality drops.

Does context budgeting work without specific external dependencies?▼

Yes, context budgeting works without external dependencies. It functions by scanning your existing Claude Code configuration files and component definitions to identify redundancy and estimate token savings natively.