Extending a Project

There's no separate add command — you extend a project by re-running the CLI inside it. The CLI reads .ai-native-project.json, skips what's already set up, and merges your new selections.

The golden rule: nothing is clobbered

• Setup files (boilerplate, Dockerfiles, config) are written only if they don't already exist — your edits are safe.
• Composed files are updated in place:
  • CLAUDE.md / AGENTS.md — new sections are appended idempotently (skipped if the section header is already present).
  • package.json — stack fragments are deep-merged, existing wins.
  • docker-compose.yml — new services are appended, with host ports chosen to avoid collisions with what's already declared.
  • CI — a job is appended per newly selected stack/app.

Common re-runs

Add a stack (single project)

Run the CLI again and pick the new stack. Its deps merge into package.json, its section/service/CI-job are appended, and its specialist agent is installed for your tools.

Add an app (monorepo)

Re-run and define another group/name/stack. Duplicate app names are rejected. The new app gets its dir, package.json, a compose service (with a free port), a CI job, and an instructions section.

Add a coding tool

Select a tool you didn't have before (e.g. add Codex to a Claude-only project). The CLI backfills that tool's layout for everything already installed — the instructions file, opencode.json, standalone and stack-bundled agents, and skills — without touching existing files. See Agentic Coding Tools.

For the assistant

The injected using-create-ai-native-project skill teaches your coding agent how to extend the project correctly (re-run to add apps/stacks/tools) so it doesn't hand-roll structure that breaks these conventions.