@runlyai/cli

Local CLI for RunlyAI Learn — the hands-on program that teaches agentic software engineering on your machine.

RunlyAI Learn is not a coding agent. Students bring Cursor, Claude Code, Codex, Copilot, or Gemini. This CLI installs the learning workflow on their project:

• Think → Plan → Build → Review → Test
• Thin runly-* routers that point at proven skills (/grill-me, /speckit-specify, /speckit-implement, /qa, …)
• Artifacts in .runly/docs/ so progress is visible and reviewable

Requirements

• Node.js 20 or newer

Installation

No global install needed. Run directly with npx:

npx @runlyai/cli

This downloads the CLI and immediately starts the setup wizard in your current project directory.

Quick Start

cd your-project
npx @runlyai/cli

The setup wizard will:

1. Scaffold .runly/ — registry, routers, STATE.md, and doc templates
2. Install .specify/ — spec-driven development infrastructure
3. Export skills to the agent folders you select (Cursor, Claude, Codex, Copilot, Gemini)
4. Create AGENTS.md if missing

Then use your coding agent with the RunlyAI Learn workflow — invoke runly-think, runly-plan, and so on from within your agent.

Commands

All commands are run via npx @runlyai/cli <command>.

init — Set up RunlyAI Learn on a project

npx @runlyai/cli init

| Option | Description | |--------|-------------| | --agent <name> | Export to a specific agent. Repeatable. Values: cursor, claude, agents (Codex & Others), copilot, gemini | | --force | Refresh templates even if .runly/ already exists | | --offline | Use bundled templates only — no GitHub fetch |

Examples:

# Default interactive setup
npx @runlyai/cli init

# Export only to Claude and Cursor
npx @runlyai/cli init --agent claude --agent cursor

# Refresh an existing setup
npx @runlyai/cli init --force

# No internet — use bundled templates
npx @runlyai/cli init --offline

export — Re-export skills to agent folders

Use this after changing which agents are in your registry.

npx @runlyai/cli export

| Option | Description | |--------|-------------| | --agent <name> | Limit export to one agent: cursor, claude, agents, copilot, gemini |

Examples:

# Re-export to all configured agents
npx @runlyai/cli export

# Re-export to Claude only
npx @runlyai/cli export --agent claude

doctor — Health check

Checks that your registry, skill files, and agent exports are all in sync.

npx @runlyai/cli doctor

| Option | Description | |--------|-------------| | --fix | Reinstall Spec Kit, sync templates from GitHub, and re-export skills | | --offline | Use bundled templates only when running with --fix |

Examples:

# Check for issues
npx @runlyai/cli doctor

# Fix issues (fetches latest from GitHub)
npx @runlyai/cli doctor --fix

# Fix issues without internet
npx @runlyai/cli doctor --fix --offline

What Gets Installed

| Piece | Location | Purpose | |-------|----------|---------| | Learn workflow | .runly/ | Registry, routers, STATE.md, doc templates | | Spec Kit | .specify/ | Spec-driven development infrastructure | | Skills export | .agents/skills/, .cursor/skills/, etc. | Routers + bundled community + Spec Kit skills |

Skills are regenerated on your machine — not committed to git.

What to Commit

.runly/
.specify/
.runly/docs/     ← PROJECT_BRIEF, SPEC, PLAN, REVIEW_NOTES, TEST_PLAN, STATE.md
AGENTS.md

Add these to .gitignore (they are regenerated by npx @runlyai/cli export):

.cursor/
.claude/
.agents/
.github/skills/
.gemini/

For Instructors & Contributors

Bundled under templates/ in this repo:

templates/
  runly/        ← registry, runly-* routers, Learn doc templates
  spec-kit/     ← .specify/ + speckit-* skills
  community/    ← matt-pocock, gstack skills (classroom-pinned)

Course content, lessons, and certification live on RunlyAI Learn (cloud). This CLI is the open-source local workflow layer students use while building real projects.

License

MIT