Wtf — Development skill for Claude Code
Explain (and optionally fix) the last shell command using Claude.
How to install Wtf
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open bendusz/wtf and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Wtf does
Explain (and optionally fix) the last shell command using Claude.
Alternatives in Development
- Prompt Net Requests — I want to enhance the AsyncPlaywrightCrawlerStrategy to optionally capture network requests and console messag 85.8k ★
- OpenAI Codex CLI — (Rust implementation) 67.8k ★
- Claude Auto Resume — A shell script utility that automatically resumes Claude CLI tasks when usage limits are lifted 732 ★
README
wtf
Explain — and optionally fix — your last shell command using Claude.
$ ls /nonexistent
ls: /nonexistent: No such file or directory
$ wtf
↻ Re-running to capture output: ls /nonexistent
The command failed because the path "/nonexistent" doesn't exist. List the
parent directory with `ls /` to see what's actually there, or create the
target with `mkdir -p /nonexistent` if that's what you intended.
Install
curl -fsSL https://raw.githubusercontent.com/bendusz/wtf/main/install.sh | bash
Restart your shell, or:
source ~/.zshrc # or ~/.bash_profile / ~/.bashrc
Pinning a version
For a real chain of custody, pin **both** `install.sh` and `wtf.sh` to the same commit SHA and verify the `wtf.sh` digest. Env vars must be applied to `bash` (after the pipe), not to `curl`:
SHA=
curl -fsSL "https://raw.githubusercontent.com/bendusz/wtf/${SHA}/install.sh" \
| WTF_BRANCH="$SHA" WTF_SHA256= bash
| Env var | Purpose |
|---|---|
WTF_BRANCH |
Ref to fetch (branch / tag / commit sha). Default: main. |
WTF_INSTALL_DIR |
Where to drop wtf.sh. Default: $HOME/.wtf. |
WTF_SHA256 |
Expected SHA-256 of wtf.sh. Aborts on mismatch. |
Usage
wtf # explain the last command
wtf --fix # also propose a fix and prompt to run it
wtf --force-rerun # re-run even if the command isn't on the safe allowlist
wtf --no-rerun # never re-run; analyze the command text alone
Short flags: `-f` (`--fix`), `-y` (`--force-rerun`), `-n` (`--no-rerun`).
`--fix` asks Claude to emit a single fix command in a fenced block. `wtf` parses it out and prompts `[y/N]` before running anything. Multi-line fix blocks are rejected instead of silently truncated.
By default `wtf` checks the last co
Related Skills
Shell Explain
Explica comandos de shell em linguagem clara. Sinaliza flags perigosas e sugere alternativas mais seguras.
0x002132 Skills Bugbounty Hunt File Upload
Hunt file upload bugs — RCE via webshell, XSS via SVG/HTML, SSRF via XXE in DOCX, path traversal via filename.
Claude Code Plus
Enhancements for Claude Code CLI: smart piped command auto-approval, shell config fix, and 850+ pre-configured
Agent Looper
Fix-until-green loop for coding agents: frozen goal, shell verify as the finish line, fresh worker each round,
Paymob Explain Error
Explain a Paymob error code, HTTP status, or error message and give the cause and fix.
Clear Explain
Explain one CLEAR-100 rule: what it catches, why it exists, the fix, and a before-and-after. Use when a findin
Related Agents
Redaction Verifier
Adversarial last check before one artifact ships - given a document claimed publish-ready (optionally with its
Harmony Spec Verify
Use this agent (normally spawned exactly once by harmony-goal in SDD Phase 5) to build, deploy and optionally
Playwright E2e Engineer
Use this agent any time the user mentions Playwright tests, E2E tests, or end-to-end tests — whether they want