Catch bugs before they hurt. ScanLine reviews your code with GPT-4 and shows you how to improve it, like having an experienced code reviewer with you 24/7.
ScanLine is an AI-based CLI tool to quickly check your commit, branch, or entire repo for:
- 🏁 race conditions
- 🔒 security gaps
- 💡 inconsistent logic
- 🚀 performance
- ☑️ consistency
- ⚙️ optimization
-
⚠️ error handling
“Scanline already saved us a few weeks on prod outages by catching race conditions” - CTO, ML infrastructure company
“I used to spend 5+ hours a day reviewing my team's code; now it's less than 1, while also uncovering more high-priority updates” - Eng Manager
“My team started complementing the quality of my PRs” - SWE
"Code gen and automatic PRs are still unreliable, but Scanline's code review is immediately useful and actionable." - Friend who's tried a lot of AI tools
For Apple M1+ Silicon (ARM64):
bash -c "$(curl -sSL https://raw.githubusercontent.com/ScanLineDev/scanline/main/install.sh)"For Apple Intel (x86):
bash -c "$(curl -sSL https://raw.githubusercontent.com/ScanLineDev/scanline/main/install_x86.sh)"then:
cd into the repo and branch to review. Run scanline.
(If you happen to not to set an OpenAI key during the install guide, then set a local .env file with your OPENAI_API_KEY, or export OPENAI_API_KEY=xxx.)
ScanLine.Demo.v0.2.0.mp4
(Click video to play)
# see all available commands
scanline --help
# check all diffed changes on your current branch compared to main or master
scanline
# only review uncommited changes on the current branch
scanline --scope commit
# review all the code in the whole repo
scanline --scope repo
# in addition you can specify a single file. For example, here's how to see the changes to the file foo.py across only this last commit
scanline --scope repo --file ./path/to/foo.py
This should not replace you doing good code review--it is meant to assist!
This is the V0 release, so it may be buggy & design may change quickly.
- Feel free to open a PR with your own changes if you'd like to see something added!
- Open an issue if you find any bugs or have an ideas
pip3 install -e .
Now you can run the CLI as reviewme above. It should auto-update if you change the python code. You may have to run pip3 install -r requirements first
Experimental: You can add rule templates to /rule_templates directory. This text will be included in the prompt to the LLM, so the LLM can evaluate the code according to the style guide you write.
Experimental: Modify config.yaml to tweak things like temperature, supported filetypes, and how many results to show per category.
Python - .py
JavaScript - .js
TypeScript - .ts
Shell, Bash - .sh, .bash
Rust - .rs
Go - .go
C - .c
C++ - .cpp, .cc, .cxx, .c++
C# - .cs
Objective-C - .m, .mm
R - .r
Ruby - .rb
PHP - .php
Java - .java
Swift - .swift
Kotlin - .kt, .kts
Scala - .scala
Perl - .pl, .pm
Lua - .lua
Groovy - .groovy, .grvy, .gy, .gvy