llm-echo 0.5 updates simulated token usage for tool calls and reasoning

llm-echo is a plugin for Willison's LLM tool that adds an echo model. The model involves no LLM at all and simply repeats back the prompts it receives. Willison revisited the plugin while testing pull requests for an upcoming LLM release.

Version 0.5 changes the simulated token counts so they take tool calls, system prompts and reasoning tokens into account. The example run uses the thinking option and the --usage flag with the prompt 'hi there', and reports 2 input and 24 output tokens.

Developers who test agent code against the echo model can now see usage figures that include tool calls, system prompts and reasoning tokens. This gives budget-tracking and logging code a closer offline test case.

The counts are simulated, so they should still be checked against real model providers before anyone relies on them for billing or limits.