Blog
- Modules in AI harnesses
How I cut my LLM API costs by splitting the system prompt and tools into modules that only load when a conversation needs them.
How I cut my LLM API costs by splitting the system prompt and tools into modules that only load when a conversation needs them.