dip is a single, self-contained binary for configuring your tenant and for building, running and
shipping data apps. The app SDK ships inside the binary, so scaffolding, running, building and pushing
an app needs no repository checkout and no package registry.
Who uses it
App developers
Scaffold, run and ship a data app with
dip app new / dev / build / push. See
Command reference.Data engineers
Connect a source, size an engine, apply its catalogs and crawl the metadata index with
dip integration, dip engine and dip crawl.Agent authors
Register the tenant’s MCP server with a coding agent, and read the business vocabulary and
metric inventory with
dip mcp, dip ontology and dip semantic.CI pipelines
Build and push an app with an API key instead of a browser login. See
Using dip in CI.
The OAuth model
dip login <workspace-url> is the only sign-in step, and the only thing you type is your
tenant’s own URL, the one in your browser’s address bar (for example
https://acme.app.us.dippr.ai). A browser window opens, you sign in the same way you would to the
console, and dip caches the resulting session on your machine. Everything else, including which
engine dip app dev talks to, is derived from that one URL — there is nothing else to configure.
Every query dip runs, whether from dip app dev or from a write command like
dip integration create, runs as you. It shows exactly the rows your own account can see and
refuses whatever your account cannot do — signing in with dip never grants you anything your
console account does not already have.
dip login requests a fixed set of capabilities for the CLI, covering the commands documented
here. If you run a command and it tells you your session does not carry a capability it needs,
running dip login again refreshes it — this is common right after an upgrade adds a new command.--workspace <url> (or the DIPPR_URL environment variable) to
say which one you mean; leave it off and a command with no ambiguity picks the one tenant you are
signed in to.
What stays in the console
A handful of decisions are deliberately console-only, because they hand out authority rather than configure something:Conventions
A dip convention is a rule that holds for every command in a group, so a script or an agent can rely on it without reading each command’s help.dip pipeline --help sums up its commands’ conventions
as: no prompts (except a push choosing a registry, below), --yes on delete, --json on list, run, runs and events, and no secret as a
flag (the owner key on standard input, an API key in DIPPR_API_KEY).
Prerequisites
dip itself needs nothing but the binary. The tools below are only needed by the commands that shell
out to them, so install what your apps actually use — and dip doctor checks all of it for you (see
Install dip and sign in).
You do not need Python installed to ship a python or streamlit app.
dip app new writes a ready
manifest.json on its own, dip app build and dip app push run the conformance check inside the
image on Python 3.12, and dip app check falls back to that same Docker check when your machine’s
python3 is older than 3.10 or absent. Host Python 3.10+ is only needed to run the app on your own
machine with dip app dev.Command map
Most commands work with either adip login session or an API key; see each command’s credential
requirements below. The authoring verbs —
dip semantic’s create/import/update/delete/propose and dip ontology’s
create/update/delete/propose/import — take a login session that consented to the authoring scopes just
as they take an API key; a session cached before those scopes existed refreshes with another
dip login. See Command reference for full flag tables.
Related
Install dip and sign in
Installing the binary and the interactive sign-in flow.
Command reference
Every
dip command’s flags, defaults and capability requirements.