It is your account, your data and your Worker. So leaving is mostly a matter of deciding to, and this page is the practical version.
The zeroth option: stop upgrading
Who owns your data is why leaving is possible at all, ejecting a capability is the partial version, and disconnecting is what ends the dashboard’s access.
The kit keeps working with nobody’s involvement.
It is MIT, it is in your node_modules, and a version that works keeps working. Nothing phones home, nothing expires, and there is no license check to fail.
If what you want is stop depending on somebody else’s release cadence, that is already available and costs nothing.
Your data is already where it is going to be
There is no export step, because there is no copy anywhere else.
Every table is in your D1, in your account. Every file is in your R2. Every secret is in your own store.
No migration is required to leave, which is unusual enough to be worth stating plainly: the thing that normally makes leaving expensive does not exist here.
Take the code you want to own
pithy add auth --ejectCopies the capability’s source into your repo and repoints the wiring at your copy. The dependencies it needed are promoted to your own manifest, so the local copy builds standalone.
Do it per capability, or for all of them. It is always available and it is one-way — no part of an ejected capability upgrades afterwards.
What you would be taking on
Worth being concrete, because fork it is easy to say.
Auth is a token model, a session store, a device registry, two tiers of rate limiting and four social providers, each with a credential shape and a linking rule.
Payments is five rails’ vocabularies mapped into nine statuses, an idempotent projection, monotonic ordering, sandbox isolation, and a reconciliation pass.
Storage is presigned uploads, resumable multipart, share revocation, quota accounting, and the untrusted-bytes response rules that stop a user’s file executing on your origin.
None of that is unreasonable to own. All of it is more than it looks like from the outside, and the honest version of we will maintain our own includes a security review each.
Leaving the hosted dashboard
Separate from leaving the kit, and simpler.
pithy dashboard disconnect --env prod --yes --localThe row goes first, and telling them is a courtesy that cannot fail — --local skips it entirely.
Every pane fails immediately, because the authority was always your Worker.
Nothing of yours was there to retrieve. What was stored is connection metadata — which projects, which environments, which addresses — and your audit trail keeps the record of everything that client ever did.
Leaving Cloudflare
This is the hard one, and the kit does not pretend otherwise.
Every capability is written against Workers, D1, KV, R2, Vectorize, Workflows and Durable Objects. There is no Postgres adapter and no container mode, and ejecting the source does not change that — the ejected code is still Workers code.
Leaving Cloudflare means a rewrite of the runtime layer, whoever wrote it. That is a real lock-in and Is Pithy for you? says so before you start rather than after.
What you keep either way is the data, in standard SQLite and standard object storage, exportable by ordinary means.
Tearing a project down
There is no teardown command for a declared environment, deliberately. A branch’s environment is disposable and has one; staging and production are not, and the one-word difference is not a difference a flag should carry.
Delete them in Cloudflare, by hand, on purpose.
If you are decommissioning: take a database export first, then delete. Nothing recomputes a name after the fact — teardown works from names derived from your project name, so changing that name first would orphan everything.
What we would rather you did
File the reason. The kit’s own dashboard filed ninety-three gaps while being built on it, and eighty-four are closed.
If you are leaving because something is missing, that is a thing that can be fixed. If you are leaving because the shape is wrong for you, that is worth knowing too — and it is on the record for the next person deciding.