Workflows

Host invocation

All sixteen workflows are first-class on both hosts:

Claude: /rn-dev-agent:<workflow> [arguments]
Codex:  $rn-dev-agent:<workflow> [request text]

Codex also has eleven domain skills, for exactly 27 native skills. Its workflow skills are explicit-only and best-effort source-command-* migration is disabled. The primary workflow is rn-feature-dev; the same names below apply to both hosts.

Command	When to use
run-workflow	Establish the proven operating sequence before any real device journey
rn-feature-dev	Build a complete feature with the 8-phase pipeline
test-feature	Test an already-implemented feature end-to-end
build-and-test	Build the app first, then test
debug-screen	Diagnose a broken screen — gather evidence, apply fix
observe	Watch the agent live in a web UI — timeline, device mirror, state panels, browser-run actions

Actions & regression

Command	When to use
list-learned-actions	List saved actions, flows, and feedback memories
run-action	Replay a saved action with self-repair
lock-e2e	Promote a verified action into a frozen, locked e2e regression test

Proof & testing

Command	When to use
proof-capture	Record video + screenshots for PR proof
nav-graph	Map and navigate the app’s screen graph

Setup & diagnostics

Command	When to use
setup	Onboard a project — Claude manages CLAUDE.md; Codex manages AGENTS.md; writes are previewed
doctor	Strictly read-only host/package/MCP/schema/task/environment diagnosis
check-env	Verify Metro, CDP, and simulator are ready
check-vercel-rules	Audit the project against the bundled best-practice rules

Utility

Command	When to use
send-feedback	Report a bug or send feedback via GitHub issue