llm-echo

llm-echo is a testing plugin developed for the LLM command-line tool, providing a fake model called echo that echoes the input prompt back in JSON format.

When using the -o thinking 1 option, the echo model first outputs a simulated reasoning block to stderr, then returns a JSON result echoing the prompt to stdout.

Before llm-echo 0.5, the echo model's token counting may have been based simply on the length of input and output text, unable to reflect complex token consumption structures.

Using the command `uvx --with llm-echo llm -m echo 'hi there' -o thinking 1 --usage` yields JSON output containing fields such as prompt, system, attachments and stream.

When using the `-o thinking 1` option, the echo model simulates outputting reasoning text, which is reflected in the token statistics as 2 input tokens and 24 output tokens.

The llm-echo plugin follows the open-source model and can be run directly via uvx without prior installation; the project source code and release notes can be viewed on GitHub.

After upgrading to llm-echo 0.5, developers can more accurately test token-usage-related logic, such as usage statistics dashboards, cost alerts and context window management, without consuming API quota.

llm-echo is a debugging plugin in the LLM command-line tool ecosystem developed by Simon Willison; it does not call a real large language model but instead returns the received prompt verbatim.

llm-echo has released version 0.5, with a focus on improving the simulated calculation of token usage.