| name | context-budget |
| description | Use when audit OpenCode context overhead across agents, skills, MCP servers, and repo instructions. Use only for workflow/context bloat decisions, not normal implementation. |
Context Budget
Adapted from ECC's context-budget skill (MIT).
Analyze token overhead across every loaded component in a OpenCode session and surface actionable optimizations to reclaim context space.
When to Use
- Session performance feels sluggish or output quality is degrading
- You've recently added many skills, agents, or MCP servers
- You want to know how much context headroom you actually have
- Planning to add more components and need to know if there's room
- Running
/context-budget command (this skill backs it)
How It Works
See REFERENCE.md for full phase details, report format, and examples.
Phases Summary
- Inventory — scan agents, skills, rules, MCP servers, AGENTS.md
- Classify — sort each component into Always/Sometimes/Rarely needed
- Detect Issues — bloat, redundancy, MCP over-subscription
- Report — produce context budget report with ranked optimizations
Quick Tips
- Token estimation:
words × 1.3 for prose, chars / 4 for code
- MCP is the biggest lever: each tool schema costs ~500 tokens
- Agent descriptions load always — even if agent is never invoked
