revenuecat-cli: driving RevenueCat from the terminal

rc is the official RevenueCat command line interface. It covers most of the same project, app, product, entitlement, offering, paywall, chart, and store-state operations as the RevenueCat MCP server (the CLI adds paywall generate/edit; the MCP server has some SDK feature-gate and experiment tools the CLI does not). Use whichever surface is available; this skill is the reference for the CLI path.

Install and authenticate

Run without installing: npx @revenuecat/cli <command> (good for CI and agent sandboxes).
Or install: brew install RevenueCat/tap/rc, or npm install -g @revenuecat/cli.
Authenticate once with rc auth login (browser OAuth), or set RC_API_KEY. Non-interactively, pass --api-key or set RC_API_KEY.
The CLI authenticates with a RevenueCat secret key (sk_...); that is correct because the CLI is a server-side tool. This is not the same as the public SDK keys (appl_/goog_/amzn_) you fetch for client app code. Never put a secret key in an app.

Discover the surface

The CLI is self-describing. Don't guess a command or flag, ask the binary:

rc commands --json lists the full command tree. Capability IDs appear in colon form (apps:create, products:store:plan).
rc commands --schemas --json returns every command's flags, args, and examples in one call. This is the most reliable way for an agent to discover per-command detail.
rc schema <command> --json returns detail for a single command, but the command must be space-separated tokens (rc schema apps create), not the colon form from rc commands. rc schema apps:create silently returns the root schema, so when in doubt use rc commands --schemas.
rc --help and rc <noun> --help give human-readable help.

Output conventions

--json gives stable, machine-readable output. Data goes to stdout, progress and chatter to stderr.
The envelope is { "data": ... }, but the shape under data varies by command: lists are usually data.items[], though some commands use other keys (charts use data.names[]). Inspect the schema or the response rather than assuming data.items.
--no-input never prompts, it fails instead. Use it in scripts, CI, and agents.
--yes / -y skips confirmation prompts on mutating commands.
--project-id <id> selects the project (or set a default with rc projects use). Pass it explicitly in scripts.
Exit codes: 0 success, 1 error, 2 bad usage, 4 auth, 5 not found, 6 rate limited.

Common operations

Look up exact flags with rc commands --schemas --json (or rc schema <space-separated command>). The common nouns:

Projects and apps: rc projects list|create|use, rc apps list|create|keys
Catalog: rc products ..., rc entitlements ..., rc offerings ..., rc packages ...
Store credentials and state: rc setup apple|google, rc products store plan|apply|sync
Data: rc charts list|show, rc customers ...
Paywalls: rc paywalls generate|edit|publish (dashboard design/audit playbook: revenuecat-paywall-design)
Dashboard: rc open [section] [id] --print (resource URLs: revenuecat-dashboard-links)

Prefer the specific command. Drop to rc api <METHOD> <path> only for endpoints not yet in the CLI surface.