wut
CLI that explains your last terminal command output using an LLM
TLDR
SYNOPSIS
wut
DESCRIPTION
wut is a terminal assistant that uses a large language model (LLM) to explain whatever was most recently printed to your terminal. It can decipher stack traces, error codes, unfamiliar log lines, and unexpected command output, returning a plain-language explanation and often a suggested next step.The tool reads the visible scrollback of the current pane via the host terminal multiplexer, so it must be invoked from inside a tmux or screen session. It sends the captured text to a configured LLM provider (e.g. OpenAI or Anthropic) using credentials stored as environment variables.
CONFIGURATION
OPENAI_API_KEY
API key used when wut is configured to call OpenAI models.ANTHROPIC_API_KEY
API key used when wut is configured to call Anthropic models.
CAVEATS
Must be run inside a tmux or screen session — wut needs the multiplexer to capture pane contents. Each invocation sends terminal output to a third-party LLM, so avoid running it on output that contains secrets, tokens, or PII. Network access and a valid API key are required.
HISTORY
wut was created by Jonathan Shobrook (shobrook) and is written in Python. It targets the same problem space as thefuck but explains errors instead of trying to auto-correct them.
