Anthropic is clearly pushing to own the scaffolding that everyone uses.
- Anthropic is clearly pushing to own the scaffolding that everyone uses.
- Claude Code is best with Max, and they'll disable use of the description in other harnesses.
- Claude Code apparently also works to thwart the KV cache of other models when used.
- But the harness has no staying power other than UI polish.
- Tokens are commodity, and for a user they are only necessary temporarily to create durable outputs.
- Once you have the outputs, you don't need more tokens.
- You only need tokens if you need to change it.
- It's trivial to make a competitor to Claude Code.
- The core agent loop is deliciously simple.
- The hardest part is the TUI but those aren't that hard given enough agentic engineering, and in some agentic harnesses you don't even need a UI.