fixr-cli 0.1.1

AI-powered CLI that explains programming errors using a persistent cache and LLMs.

fixr ⚡

AI-powered CLI that explains errors and suggests fixes — using a hashtable cache + LLM combo.

Install

pip install fixr
# or
uv add fixr

Usage

# Direct
fixr "TypeError: unsupported operand type(s) for +: 'int' and 'str'"

# Pipe any command output
python script.py 2>&1 | fixr
cargo build 2>&1 | fixr
node app.js 2>&1 | fixr
g++ main.cpp 2>&1 | fixr

Setup

# Set API key
fixr config --provider groq --api-key gsk_...

# Set default provider
fixr config --provider gemini

# OAuth login (Google/Gemini)
fixr login google

# Show current config
fixr config --show

# List all providers
fixr providers

Free tier providers (no credit card)

Provider	Models
groq	llama-3.3-70b-versatile
gemini	gemini-2.0-flash
mistral	mistral-small
openrouter	20+ free models
nvidia	100+ open models
cerebras	llama3.3-70b

How it works

error input
    │
    ▼
normalize + hash (sha256[:16])
    │
    ▼
cache hit? ──yes──→ instant response ⚡
    │
   no
    ▼
LLM call (provider of choice)
    │
    ▼
store in ~/.fixr/cache.json
    │
    ▼
display fix

Cache lives at ~/.fixr/cache.json. Gets smarter over time.