← ClaudeAtlas

context-budgetlisted

Token and context budget audit for agents, skills, rules, and MCP overhead. Use when sessions feel bloated or repeated references are wasting context.
JunMystery/AI-Agent-Standards · ★ 0 · AI & Automation · score 73
Install: claude install-skill JunMystery/AI-Agent-Standards
# Context Budget Use this skill to reduce wasted context and identify bloat. ## Trigger - Session quality degrades as context grows - More agents, skills, or rules were added recently - You want to know what should remain on-demand ## What to Apply - Prefer on-demand load over always-loaded instructions. - Identify overlapping docs and duplicate guidance. - Keep root instruction files short. - Move specialized guidance into task-specific files. ## Reference Files - [SKILL-REFERENCE.md](../../SKILL-REFERENCE.md) - [README.md](../../README.md) - [ai-agent-standards/INDEX.md](../../ai-agent-standards/INDEX.md) ## Output Expectation Report the biggest token savings first and separate always-loaded from on-demand content.