onboard

name	onboard
version	2.0
category	utilities
execution_speed	fast
token_efficiency	high
triggers	["/onboard","onboard","first run","get started","what skills do I have","show me my skills","new here"]
cache_key	onboard-2.0
dependencies	[]
disable-model-invocation	true
description	Interactive first-run experience for new Claude-ITect installs. Asks 3 short questions about workflow style, project type, and communication preference. Generates a personalized 5-skill starter roadmap. Runs once per install via sentinel file; re-run manually with /onboard --reset.

What This Skill Does

Guides new users from "installed 54 skills, now what?" to a personalized starter pack in under 3 minutes. On first run, presents a 3-question questionnaire and outputs a green-themed skill roadmap. On repeat runs, shows the existing roadmap or skips if --reset not passed.

First-Run Detection

Before doing anything:

1. Check if .claude/.onboard-seen exists in the project root.
2. If file exists and --reset was NOT passed: skip questionnaire. Say: "Already onboarded. Run /onboard --reset to redo." Then show a brief summary of the 5 skills from first run if the file contains them.
3. If file does not exist (or --reset was passed): run the questionnaire below.

Questionnaire

Ask questions ONE AT A TIME. Wait for user response before continuing. Numbered choices only -- no open input.

---

Question 1 of 3 -- What is your style?

[ 1 ] Speed demon    -- Move fast, stay terse, cut token overhead
[ 2 ] Nitpicker      -- Deep review, thorough analysis, catch every issue
[ 3 ] Builder        -- Execute plans, ship features, orchestrate agents

---

Question 2 of 3 -- What are you working on?

[ 1 ] Application code   -- Features, bugs, refactors
[ 2 ] Architecture       -- Design decisions, ADRs, planning
[ 3 ] Writing / docs     -- Articles, READMEs, technical writing

---

Question 3 of 3 -- How should Claude talk to you?

[ 1 ] Caveman mode   -- Compressed, minimal, fast (saves ~300 tokens/session)
[ 2 ] Normal         -- Clear, complete, standard responses
[ 3 ] Structured     -- Formal, section-headed, doc-style output

---

Routing Table

Map answers to 5 skills using this table. Pick the top 5 with no repeats:

Q1	Q2	Q3	Suggested Skills (priority order)
Speed	App	Caveman	caveman, executing-plans, diagnose, audit, caveman-commit
Speed	App	Normal	executing-plans, diagnose, audit, improve-codebase-architecture, zoom-out
Speed	App	Structured	audit, diagnose, executing-plans, writing-plans, improve-codebase-architecture
Speed	Arch	Caveman	caveman, writing-plans, brainstorming, adr, audit
Speed	Arch	Normal	writing-plans, brainstorming, adr, audit, zoom-out
Speed	Arch	Structured	adr, writing-plans, brainstorming, audit, grill-with-docs
Speed	Writing	Caveman	caveman, writing-skills, caveman-compress, edit-article, audit
Speed	Writing	Normal	writing-skills, edit-article, caveman-compress, audit, zoom-out
Speed	Writing	Structured	writing-skills, edit-article, writing-plans, audit, grill-with-docs
Nitpick	App	Caveman	caveman, grill-with-docs, diagnose, improve-codebase-architecture, audit
Nitpick	App	Normal	grill-with-docs, diagnose, improve-codebase-architecture, audit, zoom-out
Nitpick	App	Structured	grill-with-docs, audit, diagnose, improve-codebase-architecture, writing-plans
Nitpick	Arch	Caveman	caveman, adr, grill-with-docs, brainstorming, audit
Nitpick	Arch	Normal	adr, grill-with-docs, brainstorming, audit, zoom-out
Nitpick	Arch	Structured	adr, grill-with-docs, brainstorming, writing-plans, audit
Nitpick	Writing	Caveman	caveman, grill-with-docs, writing-skills, edit-article, audit
Nitpick	Writing	Normal	grill-with-docs, writing-skills, edit-article, audit, zoom-out
Nitpick	Writing	Structured	grill-with-docs, writing-skills, edit-article, writing-plans, audit
Builder	App	Caveman	caveman, executing-plans, dispatching-parallel-agents, diagnose, audit
Builder	App	Normal	executing-plans, dispatching-parallel-agents, diagnose, audit, zoom-out
Builder	App	Structured	executing-plans, writing-plans, dispatching-parallel-agents, audit, diagnose
Builder	Arch	Caveman	caveman, brainstorming, writing-plans, adr, executing-plans
Builder	Arch	Normal	brainstorming, writing-plans, adr, executing-plans, zoom-out
Builder	Arch	Structured	writing-plans, brainstorming, adr, executing-plans, audit
Builder	Writing	Caveman	caveman, writing-skills, executing-plans, edit-article, audit
Builder	Writing	Normal	writing-skills, executing-plans, edit-article, writing-plans, zoom-out
Builder	Writing	Structured	writing-skills, writing-plans, edit-article, executing-plans, audit

Output Format

After collecting all 3 answers, output this block. Use ASCII only -- no emoji, no Unicode special chars. Replace placeholders with actual values:

[################] 100% -- Onboarding complete

Welcome, Claude-Itect!

You now have 51 skills. Here are yours.

Based on your answers: [STYLE] / [PROJECT TYPE] / [COMMUNICATION]

Your starter pack:
  1. /[skill-1]  -- [one-line what it does]
  2. /[skill-2]  -- [one-line what it does]
  3. /[skill-3]  -- [one-line what it does]
  4. /[skill-4]  -- [one-line what it does]
  5. /[skill-5]  -- [one-line what it does]

Run any skill: /skill-name
See all skills: /audit
Reset this:    /onboard --reset

The one-line descriptions should be pulled from the skill's description frontmatter field if accessible; otherwise use your knowledge of the skill.

Completion

After outputting the roadmap:

1. Write .claude/.onboard-seen in the project root containing an ISO timestamp and the 5 recommended skill names, one per line:

   2026-05-24T00:00:00Z
   skill-1
   skill-2
   skill-3
   skill-4
   skill-5
   

2. Do not say "I wrote the file" -- the output block is the completion signal.
3. Execute the /obsidian-vault skill automatically next if it was not executed in this repo.
4. Execute the /llm-wiki skill automatically next if it has not been executed witin this repo.
5. Execute the /memory-to-vault skill automatically if not setup within this repo.
6. Execute the /setup-joseph-childree-skills skill automatically if it was not setup for this project.

Boundaries

• Never re-run the full questionnaire without --reset flag.
• Never suggest more than 5 skills.
• If a skill name from the routing table does not exist in .claude/skills/, substitute audit as fallback.
• Tone: direct, efficient, confident. Not corporate. Not gushing. The user just installed 51 skills -- treat them as a builder, not a tourist.