How-to
How to bring an existing repo under management
Import the whole repo in one shot. On first connect, CatWrangler reads your existing code and docs, extracts the decisions behind them, and hands your agents a pre-digested briefing. From there you build normally — describe a change, get the files, ship.
An existing codebase isn't a blank slate, and CatWrangler doesn't treat it like one. You bring the repo in whole; the system reads what's already there so your agents start informed, not lost.
What happens when a repo comes in
Most tools make you onboard a codebase by crawling it — agent after agent grepping its way to a mental model that evaporates between sessions. CatWrangler reads your existing code and docs once, on first connect, and turns them into something durable: the decisions behind the work, bound to the exact code, ready for any agent to read before it builds.
It works on large and legacy codebases — the messy ones with years of context nobody wrote down. That context stops being lost the moment the repo is under management.
- Bring a repo in (or take it back out) in one shot — no lock-in, nothing to self-host.
- Existing code and docs are read on first connect, so agents don't start from zero.
- Decisions are extracted from what's already there — the 'why' becomes a first-class artifact.
- Once imported, you build normally: describe a change, get the files and the decisions behind them, ship.
- 01
Import the repo in one shot
Point CatWrangler at your existing repository and bring it in whole — no piecemeal migration, no learning version control. It runs as a hosted service on your own isolated server, so there's nothing to set up and nothing to self-host. If you ever want out, you take the repo back out the same way: one shot, no lock-in.
- 02
Let it read what you already have
On first connect, CatWrangler reads your existing code and documentation. This is the whole point of importing rather than starting fresh — it works on large and legacy codebases, so even years of accumulated context becomes something the system understands instead of something your agents have to rediscover.
- 03
Get the decisions extracted
From the code and docs it just read, CatWrangler surfaces the decisions behind the work — the choices, the reasoning, the alternatives — and binds them to the exact code they govern. The 'why' that was scattered across commits, chat logs, and people's heads becomes a structured artifact your agents can read.
- 04
Brief your agents instead of making them crawl
Now an agent describes its intent and gets a pre-digested briefing back: the project, its structure, the relevant decisions, and where to work. No grepping the repo to build a model that disappears at the end of the session. Every agent and sub-agent gets its own identity and line of work instantly, with zero setup.
- 05
Build normally from here
With the repo under management, you work the way you would on any CatWrangler project. Describe a change in plain language and get the exact files and functions plus the decisions behind them. Before building, the system surfaces code and decisions that already do the job — so an agent extends rather than duplicates, often finding the work was already done.
- 06
Ship it
Submit a change and it's validated — it must cover the decision it claims and build and pass tests before it merges. Then it builds, validates, and deploys itself, with no separate pipeline. A bad deploy rolls back; a tagged prior state can be restored. Works for any language or project type.
“Reclaimed my identity, branch preserved. Trunk has moved on — others have been shipping. Zero active conflicts, nothing waiting on me.”
Related questions
Does my repo have to be small or new?
No. Onboarding is built for large and legacy codebases. On first connect, CatWrangler reads your existing code and docs so your agents start with real context instead of a blank slate — the older and messier the repo, the more there is to recover.
Am I locked in once I import?
No. You bring a repo in or take it back out in one shot. Real version control runs underneath the whole time; you and your agents never see it, and you're free to leave with your repo whenever you want.
Keep reading
Vibe-Engineering
Many agents. One codebase. Zero collisions.
Point your agents at CatWrangler and build — the discipline runs underneath.