ai-new-project

name	ai-new-project
description	Stand up a new repository with standard quality gates, CI/CD pipeline, and configuration files.
argument-hint	[repo-name]

Create a new standardized repository: $ARGUMENTS

Tool-first project bootstrap. Use ai-shell for scaffolding and ai-tools for deterministic workflow and standards behavior.

1. Gather Inputs

Collect:

• repo name (use $ARGUMENTS when provided)
• repo type: library or service (use workspace only when intentionally creating a coordinator repo)
• language/ecosystem
• framework (for services)
• visibility (private or public)

2. Create the Repository

Use gh to create and clone the repo.

For service repos, create and push a development branch (dev/develop/staging) based on team convention.

3. Bootstrap Local Project Files

Create the baseline project skeleton for the selected ecosystem, then run package manager initialization commands.

Do not manually encode the full standards policy in this skill. Keep local scaffolding minimal and delegate standards rules to ai-tools standardize.

4. Scaffold AI Tooling

Run ai-shell initialization with the selected repo type and target tools:

ai-shell init --library
ai-shell init --service
ai-shell init --workspace

5. Apply Standards via ai-tools standardize

Run:

ai-tools standardize detect --json
ai-tools standardize audit --json
ai-tools standardize fix --write --json
ai-tools standardize verify --json

If the user asks for audit-only mode, skip fix and stop after audit.

6. Verify Workflow Surface

For normal repos, confirm:

• ai-tools repo status --json works
• ai-tools repo branch prepare --json is the branch-prep entrypoint
• ai-tools repo submit --json is the submit entrypoint

For workspace repos, confirm:

• ai-tools mono status --json works
• workspace skills map to ai-tools mono ...

7. Initial Commit and Push

Commit the scaffold with a conventional commit and push to remote.

For service repos, ensure branch strategy is explicit (dev -> main) before starting feature work.

Error Handling

• gh not authenticated: ask user to run gh auth login.
• ai-tools missing commands: report exact missing subcommand and continue with the closest available audit/status command.
• Network failures: retry once, then report and continue with local setup where possible.