/getting-started
Guided five-minute onboarding for Director Mode Lite across Claude Code, Codex CLI, and Grok Build. Use after installation or when choosing a first workflow.
$ npx -y skills add claude-world/director-mode-lite --skill getting-started --agent claude-codeHow it fires
How this skill gets triggered: by you, by Claude, or both.
- Fires itselfAuto-invocation. Claude auto-loads it when your prompt matches the work.
- You can call itInvoke it directly when you want it.
- Slash command
/getting-started
Context preview
The summary Claude sees to decide when to auto-load this skill.
Guided five-minute onboarding for Director Mode Lite across Claude Code, Codex CLI, and Grok Build. Use after installation or when choosing a first workflow.
SKILL.md
getting-started.SKILL.mdname: getting-started
description: Guided five-minute onboarding for Director Mode Lite across Claude Code, Codex CLI, and Grok Build. Use after installation or when choosing a first workflow.
user-invocable: true
Getting started
Director Mode is a set of adaptable workflows, not a permission or policy layer. Keep the user's preferred CLI controls and choose only the skills useful for the current task.
1. Verify the portable surfaces
./scripts/verify-install.sh /path/to/project
A full three-CLI install currently provides 35 skills, 14 canonical Claude agents, 14 Codex adapters, 14 Grok adapters, shared guidance, and the session relay. Python 3 is required for installation and relay packets. `jq` is only a legacy automation dependency; a normal zero-hook install does not need it.
Active hooks are optional. The default install deliberately registers none. Use `--hooks guide` only when a short non-blocking Claude/Codex SessionStart context is useful. Grok reads the same guidance from `AGENTS.md` without an inert passive hook.
2. Choose a starting point
- `director-mode`: define outcome, context, constraints, evidence, and the next
decision for substantial work.
- `workflow`: use the full research → plan → implement → verify → review flow.
- `focus-problem`: understand an unfamiliar bug or codebase area.
- `session-relay`: leave portable state for Claude, Codex, or Grok to continue.
- `smart-commit`: review and prepare a conventional commit when requested.
Testing and TDD skills are available when they fit the repository or the user asks for them; they are not imposed on every task.
3. Let another CLI continue
.director-mode/bin/director-relay create \
--from claude --to codex \
--goal "Finish the current feature" \
--summary "Implementation is partly complete" \
--next "Inspect the worktree and continue"
.director-mode/bin/director-relay continue --to codex
The second command prints a copyable native command. It launches nothing until the user explicitly adds `--run`.
Help
- `agents`: list available agent roles.
- `skills`: browse the skill catalog.
- `docs/FAQ.md`: installation and relay questions.
- <https://claude-world.com>: tutorials and updates.
Read more
name: getting-started description: Guided five-minute onboarding for Director Mode Lite across Claude Code, Codex CLI, and Grok Build. Use after installation or when choosing a first workflow. user-invocable: true
Getting started
Director Mode is a set of adaptable workflows, not a permission or policy layer. Keep the user's preferred CLI controls and choose only the skills useful for the current task.
1. Verify the portable surfaces
./scripts/verify-install.sh /path/to/project
A full three-CLI install currently provides 35 skills, 14 canonical Claude agents, 14 Codex adapters, 14 Grok adapters, shared guidance, and the session relay. Python 3 is required for installation and relay packets. `jq` is only a legacy automation dependency; a normal zero-hook install does not need it.
Active hooks are optional. The default install deliberately registers none. Use `--hooks guide` only when a short non-blocking Claude/Codex SessionStart context is useful. Grok reads the same guidance from `AGENTS.md` without an inert passive hook.
2. Choose a starting point
- `director-mode`: define outcome, context, constraints, evidence, and the next
decision for substantial work.
- `workflow`: use the full research → plan → implement → verify → review flow.
- `focus-problem`: understand an unfamiliar bug or codebase area.
- `session-relay`: leave portable state for Claude, Codex, or Grok to continue.
- `smart-commit`: review and prepare a conventional commit when requested.
Testing and TDD skills are available when they fit the repository or the user asks for them; they are not imposed on every task.
3. Let another CLI continue
.director-mode/bin/director-relay create \ --from claude --to codex \ --goal "Finish the current feature" \ --summary "Implementation is partly complete" \ --next "Inspect the worktree and continue" .director-mode/bin/director-relay continue --to codex
The second command prints a copyable native command. It launches nothing until the user explicitly adds `--run`.
Help
- `agents`: list available agent roles.
- `skills`: browse the skill catalog.
- `docs/FAQ.md`: installation and relay questions.
- <https://claude-world.com>: tutorials and updates.
Use Claude Code like a Director, not a Programmer. MIT toolkit with Auto-Loop, guided setup, 27 commands, 14 agents, and 32 skills.
Other skills on director-mode-lite.
- /changelog-observer
Track development session events in a daily markdown changelog, including file changes, test results, and key decisions.
Open skill - /agent-check
Validate custom agent file format and structure. Use after creating or editing an agent, before committing agent changes, or when an agent fails to load.
Open skill - /agent-template
Generate custom agent from template. Use when creating a new subagent from scratch, or scaffolding an agent file with correct frontmatter.
Open skill - /agents
List all available agents (core, expert, self-evolving). Use when the user asks what agents are available or runs /agents.
Open skill - /auto-loop
TDD-based autonomous development loop with checkpoint recovery and observability changelog
Open skill - /changelog
View and manage the runtime changelog for observability
Open skill

