Skip to content
  1. Home
  2. Agent skills
  3. context-budget

context-budget

Routes to the live context meter; audits installed components on request

What it is

A skill from ECC that answers context-window questions by routing to the harness's own meter (/context in Claude Code, /status in Codex) without scanning files. Only when explicitly asked, it runs a bounded static audit of installed agents, skills, MCP servers and rules to find what consumes context.

Good for

  • Check remaining context in a session
  • Find which skills or MCP servers bloat context
  • Plan which components to remove
Worth knowingThe static audit gives estimates, not exact token counts.

Sources

The summary is ours, written from the README. Spotted an error? See how we work.

More for coding

See the topic →
  • diagnosing-bugs

    mattpocock/skills

    Debugging discipline that starts by building a tight pass/fail feedback loop

    SkillGitHub stars: 323k
  • Skills For Real Engineers

    mattpocock/skills

    Small composable engineering skills from Matt Pocock's own agent setup

    SkillGitHub stars: 323k
  • Superpowers

    obra/superpowers

    Skill set that makes coding agents spec, plan, test and review before coding

    SkillGitHub stars: 317k
  • Systematic Debugging (Superpowers)

    obra/superpowers

    Forces root-cause investigation before an agent proposes any fix

    SkillGitHub stars: 317k