Workshop

$ open workshop/explain-error

● AI mode · text goes to a Claude model · daily limit

Error Explainer

Any error in plain words, plus a fix prompt

Red text in a terminal is scarier than it deserves to be. Most of the time it is one of a dozen ordinary failures: a package not installed, a wrong file path, a missing environment variable, a busy port, an outdated version. Here an AI reads what you paste and answers in four steps: what happened in plain words, the most likely cause, what to check with your own eyes, and a ready prompt for Claude Code. The prompt asks the agent to confirm the cause and show the change first, and to apply it only once you agree. The explanation itself runs nothing and changes nothing in your project. The text goes to our server and a Claude model, so strip keys and passwords from the log beforehand; if any slip through, they will not appear in the answer.

$ explain --error

● checking your limit…

0 / 6,000 characters
Ctrl + Enter

Explanation

You will get a plain-words explanation, the likely cause, what to check, and a ready prompt to paste into Claude Code. Remove keys and passwords from the log before pasting.

What this error means and how to ask Claude Code to fix it

Paste the error or a piece of the log with the lines around it and press Explain the error. In a few seconds you get four parts: what happened in plain words, the most likely cause and a couple of alternatives, what to check without risk, and a ready prompt for Claude Code. The prompt asks the agent to confirm the cause and show the change first, and to apply it only once you agree.

How to use it

  1. 1Copy the whole error, with a few log lines around it
  2. 2Paste it into the field and press Explain the error
  3. 3Read the explanation, copy the prompt with one click, and paste it into Claude Code

The details

What you get
Explanation, likely cause, what to check, a prompt for Claude Code
Input length
Up to 6,000 characters
Runs per day
Guest 2, free account 3, full access 20
Explaining the same text again
Free, from cache
Model
Claude Haiku 4.5
What it runs
Nothing: it only reads the text and answers

Common questions

Will it fix the error for me?

No, on purpose. The explainer only explains and writes the prompt. Claude Code does the fixing in your project: the prompt asks it to check the cause against your files, show the change, and apply it only after you say yes. You see what changes, and nothing breaks silently.

Can I paste a log that has keys or passwords in it?

Better not: the text goes to our server and a Claude model. Replace keys, tokens and connection strings with something like XXX before pasting. If a secret does slip in, it will not appear in the answer, and the explanation will advise you to rotate it.

Why is the number of runs per day limited?

Every explanation costs us a model call. AI mode needs an account: a free one gets 3 runs in total, full access 20 a day. Explaining exactly the same text again does not use a run: the answer comes from the cache.

Which errors does it understand?

Any text from a terminal, a browser or an agent's reply: Next.js and Node build errors, npm and pip failures, Python tracebacks, Git, Docker and Vercel messages, and Claude Code's own replies. The more lines around the error you paste, the sharper the cause.