Why Painted Wolf Code
Your machine. Your models. Your rules.
Not a chat box bolted to an editor. A coordinator puts a team of specialists on your codebase, cites the files behind every claim, and works inside limits the operating system enforces. Built from scratch, open source, on your own hardware and keys.
Pay for tokens, not a middleman.
Most tools wedge themselves between you and the model and bill you for it. We connect straight to the providers you want, cloud or local. Fan your agents across all of them at once: no single rate limit, just raw parallel speed. Cheaper, better, faster.
A whole eng team at your disposal.
A coordinator runs the show and hands work to specialists: an implementer, a path explorer, a code reviewer, a plan writer, a skeptic, a test writer, a web researcher. They work in parallel, each capped at the tools its role allows, on the models you assign — your strongest for the coordinator, something cheaper for the workers. You steer a team instead of babysitting one long thread.
Limits the model can't talk its way past.
On macOS, every command runs as a program and its arguments — never a shell string — inside its own operating-system sandbox, with your keys and the app’s own settings out of reach. That boundary decides when you get asked: contained work just runs; crossing it, or reaching into a cloud account, stops for approval. Real, deterministic scanners run alongside, and their findings reach the agent as evidence it can’t hand-wave.
Receipts, not vibes.
Every answer cites the exact files and lines behind it, checked against what the agent actually read. Shaky claims have nowhere to hide. You review the evidence, not just the conclusion.
An AI-first code editor.
Files closes the loop between asking, changing, and reviewing. Citations land at the exact line; select any code to turn it into an instruction; then watch agent edits arrive in the file while your own draft stays intact. Shape the work in the same place you inspect it, without copy-pasting context or handing work off to another tool.
Quick fix now. Full plan when it counts.
Open it and type. No ceremony for everyday work. When the change is bigger, flip on a structured flow from research to plan to build to review, with stops where you approve the next move. Need to keep one chat separate? Give it its own Git branch and checkout, then land the committed result when you are ready. Every workflow and persona — along with much of the guidance and policy they follow — is YAML in a pack you can fork. New packs install from a git URL.
Built to work differently
Ready to run it on your machine?
Download the app, connect your model provider, and start coding. Docs are there when you want them.