5 entries
A local proxy (RTK) shows 4.4M "saved" tokens (45.3%) across 9,892 cumulative commands (global scope). That figure reads like a lower bill; that reading is incomplete. That "saving" only exists because a harness (hooks, gates, routing) produces it and keeps producing it. A deliberate share of my AI budget has to fund that upkeep, not just more generation tokens.
A counter built into the tool can't measure the tool's savings: it sees neither the counterfactual, nor the cache, nor retries. Measured against my invoice, the real gain of four compression tools lands between 0 and 15%, while cache reads, meanwhile, make up 97% of tokens and re-read every token ~36 times.
Anthropic is running the Apple playbook: control the whole chain, from model to terminal. I left Claude Code for multi-provider freedom and came back for the ecosystem — not the model. The lock-in is real, but nobody else ships anything close to an OS for agentic development.
A rule written in a CLAUDE.md, the agent eventually forgets: a few tens of thousands of tokens later, it's no longer in its window. A guardrail — a hook that refuses the action — doesn't forget. I moved the rules that matter from markdown to about fifteen hooks. Here's what it changes, and what it really costs.
Claude Code is an excellent starting point — zero config, immediate quality. But Anthropic lock-in prevents optimizing costs per task, and restricting the Claude Code subscription in third-party tools forced the migration. OpenCode then Kilo Code CLI (its fork with free gateway) solve both without sacrificing quality.