AI-powered terminal error summarizer
When a command fails, clifr instantly explains what went wrong and gives you the exact fix — in plain English, before you've finished reading line one.
$ git push origin fake-branch
fatal: 'fake-branch' does not appear to be a git repository
────────────────────────────────────────────────────────
what remote branch doesn't exist yet
fix git push --set-upstream origin fake-branch
────────────────────────────────────────────────────────
$ pip3 install tensorflow
ERROR: Could not find a version that satisfies the requirement tensorflow (from versions: none)
ERROR: No matching distribution found for tensorflow
────────────────────────────────────────────────────────
what tensorflow doesn't support Python 3.14 yet
fix python3.13 -m pip install tensorflow
tip tensorflow supports up to Python 3.13
────────────────────────────────────────────────────────
- runs your command normally — output prints immediately, zero delay
- if it fails, explains the error using an LLM
- self-routing: fast
llama-3.1-8bhandles simple errors; escalates tollama-3.3-70bwhen it flags the problem as too complex - live registry lookups: pip errors hit PyPI, npm errors hit the npm registry, brew hits Homebrew, gem hits RubyGems — real version data, not hallucinated fixes
- fix command is auto-copied to clipboard
- confidence bar only shown when the model is uncertain (< 60%)
- zero dependencies — uses Node's built-in
fetch, nonode_modules
requirements: Node.js ≥ 18, macOS
git clone https://github.com/utsav-develops/clifr
cd clifr
npm linkget a free Groq API key at console.groq.com — no credit card needed.
ep --setup
# or
clifr --setupboth ep and clifr are the same command — use whichever you prefer.
wrap any command with ep or clifr:
ep git push origin main
ep npm install
ep pip3 install tensorflow
ep brew install node
ep gem install railsinstall once, and every failing command in your terminal gets explained automatically. no prefix needed.
ep --install-hook
source ~/.zshrc # or ~/.bashrcafter that, just use your terminal normally:
$ pip3 install tensorflow
ERROR: No matching distribution found for tensorflow
────────────────────────────────────────────────────────
what tensorflow doesn't support Python 3.14 yet
fix python3.13 -m pip install tensorflow
────────────────────────────────────────────────────────
| command | description |
|---|---|
ep --setup |
save your Groq API key |
ep --install-hook |
auto-explain all failing commands in your shell |
ep --raw <cmd> |
run a command without explanation |
ep --version |
show version |
| env var | description |
|---|---|
CLIFR_OFF=1 |
temporarily disable the shell hook |
CLIFR_DEBUG=1 |
show registry lookups and raw model responses |
| ecosystem | registry | what clifr fetches |
|---|---|---|
| Python / pip | PyPI | latest version, requires_python, tested Python versions |
| Node / npm | npm registry | latest version, engine requirements |
| Homebrew | formulae.brew.sh | formula version, description |
| Ruby / gem | RubyGems | gem version, Ruby requirement |
Groq's hardware is built for inference speed. llama-3.1-8b-instant has a time-to-first-token of ~200ms on short prompts. For most errors the explanation appears almost immediately after the command output. The 70b model only kicks in when the 8b flags an error as too complex (version conflicts, dependency issues, etc.).
see CONTRIBUTING.md
MIT © Utsav Acharya