Cutting cost and context waste in multi-agent Claude workflows
Claude Code can waste tokens and money when several agents run for a long time or reread the same material. Long-lived agents may still create extra cost even when caching is used, so the cache tax needs to be understood and reduced.
Subagents should receive only the exact context they need instead of rereading broad project material again and again. Important routing decisions should not depend only on a weaker model’s judgment for complex choices.
Clear outside structures, such as GitHub labels or issues, can make routing more reliable. The main idea is to design cost control, , and routing as separate parts of the .
Key points
- Long-lived agents can create a cache tax that raises cost.
- Subagents waste tokens when they reread the same broad context repeatedly.
- Passing targeted context to subagents can reduce duplicate reading.
- Routing is more reliable when tied to GitHub labels or issues instead of only model judgment.
- Cost, context, and routing should be designed as separate parts of a .