Bash Command Explainer
AI-assisted bash explanation. 5 free per day.
AI-assisted explanations are usually right but not guaranteed. Never run a command you do not understand on a system you care about, even if the explanation looks reasonable.
What this tool does
Reads a bash command (or a short script) and produces a plain-English walkthrough: what each command, flag, pipe, redirection, and substitution does, plus a notes section if the snippet has uncommon constructs or likely safety issues. Useful for reading commands from blog posts, gist links, or teammates' scripts before running them blind.
How to use it
Paste the command. Press Explain. The free tier allows 5
explanations per IP per day. Try a representative example:
find . -type f -name "*.log" -mtime +30 -exec rm \;.
Common use cases
- Reading a one-liner from a Stack Overflow answer before running it.
- Understanding a legacy shell script in a codebase you inherited.
- Explaining a complex pipe chain to a teammate without typing it out.
- Spotting an obviously-dangerous construct (rm -rf, curl-pipe-to-bash) before pasting.
- Decoding a command copied from a tutorial that does not explain what each flag does.
Common pitfalls
- Vendor differences. bash, dash (Debian sh), zsh, and POSIX sh differ in subtle ways. Most commands work the same; some constructs (process substitution, [[ ... ]], arrays) are bash-only. The explainer assumes bash.
- Environment matters. A command that uses
$HOME,$PATH, or$0behaves differently across shells, users, and contexts. The explanation describes the syntax; the runtime depends on your environment. - Always-run-as-root warnings. A command that requires root may be safe in admin context and disastrous in user context (or vice-versa). The explainer flags suspicious patterns but does not know your specific risk model.
Frequently asked questions
- Where does my command go?
- The command is sent to glunty which forwards it to Anthropic Claude. glunty does not log or store; Anthropic processes for the duration of the request under their data-usage policy (no training on API inputs by default). Strip credentials, API keys, file paths to private data, and secrets out of the command before pasting. A command that exposes sensitive infrastructure should not leave your machine.
- Will it warn me about dangerous commands?
- Often yes. Patterns the model recognizes as potentially destructive (rm -rf /, curl piped to bash, dd of=/dev/sda, mass chmod 777) get flagged in the notes section. The warnings are heuristic; the model can miss something genuinely dangerous or flag something innocuous. Treat all command warnings as "look closer," not "this is fine" or "this is doomed." Never run a command you do not understand on a system you care about.
- Does it handle complex pipes and redirections?
- Yes. Pipes (|), file redirection (>, >>, <, 2>&1), here-strings (<<<), here-docs (<<EOF), process substitution (<(...)), and command substitution ($(...) or backticks) are all explained in order. For multi-line scripts, the explanation walks line by line.
- What about non-bash shells (zsh, fish, dash)?
- bash is assumed by default. Most commands work the same in zsh and dash; some constructs are bash-only ([[ ... ]], arrays, process substitution). fish has very different syntax. The explanation calls out bash-specific constructs but does not always identify which non-bash shell variants reject them. If you use fish, expect more dialect-specific surprises.
- Why 5 per day?
- Cost ceiling for the free tier. Bash one-liners are short, so each call is cheap, but allowing unlimited use would invite automation abuse that would force a paywall. Five is enough for most casual use; for higher volume, the Claude API directly costs about a cent or two per explanation.
- Can it explain what a command WOULD do without running it?
- Yes; that is the whole point. Pasting a command from a Stack Overflow answer or a tutorial and reading the explanation before running it is much safer than running it blind. The explanation describes effects (creates files, deletes files, sets permissions, sends network requests) without those effects actually happening on your machine.
Cite this tool
For academic, journalistic, or technical references. Pick a format:
Citations use 2026 as the publication year. Access date is left as a fillable placeholder where the citation style expects one.