New: 97,000+ skills indexed from GitHub repos with 100+ stars · every skill safety-checkedBrowse top skillsTop skills
Esc to closeBrowse all skills
Submit a skillBrowse skills
Submit a skillBrowse skills

Agent skill

Personal Assistant Onboarding

by mp-web3 in mp-web3/claude-starter-kit

Walks a new user through a five-step first session, profile, problems, goals and initial tasks, to turn the agent into a personal assistant that remembers them.

MITAuto-check: notesAgent Workflows

Install Personal Assistant Onboarding

skills CLI
$ npx skills add mp-web3/claude-starter-kit --skill onboard -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install mp-web3/claude-starter-kit onboard --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/mp-web3/claude-starter-kit.git skills-src && mkdir -p .claude/skills && cp -r skills-src/skills/onboard .claude/skills/onboard && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
onboard
GitHub stars
109
Token cost
~2.7k tokens
SKILL.md length
1,205 words
Files
1
Skills in repo
6
Repo updated
First seen
Licence
MIT

At a glance

Walks a new user through a five-step first session, profile, problems, goals and initial tasks, to turn the agent into a personal assistant that remembers them.

  • Works in 5 steps: User Profile (5 min) → 12 Favorite Problems (10 min) → End Goal and Subgoals (5 min) → …
  • Running the first session after installing a personal-assistant starter kit
  • SKILL.md covers Before Starting, Step 1: User Profile (5 min), Step 2: 12 Favorite Problems… and Step 3: End Goal and Subgoals…, plus 4 more sections
  • Calls git, python3 and gh

What it does

The skill runs a guided, conversational setup meant for the very first session after installing a personal-assistant starter kit. It reads the existing CLAUDE.md and any rule or knowledge files first, then checks a progress file to resume from the last completed step if the user disconnected partway through, telling them which step they left off on. After every single step it writes the step number and the files touched into that progress file and makes a local git commit, with no push, since a remote may not be configured yet, so no step's work is ever lost to batching.

Step one builds knowledge/user/profile.md through an open conversation that asks one question at a time rather than a list, drawing out career, current projects, communication style and working preferences such as IDE and OS. The excerpt breaks off while still gathering this first step, before the problems, goals, subgoals and initial-tasks steps it promises.

When your agent uses it

  • Running the first session after installing a personal-assistant starter kit
  • Resuming an onboarding conversation that was interrupted partway through
  • Setting up a user profile, goals and initial tasks for a new assistant

Example prompts

  • “Run /onboard to set up my personal assistant for the first time.”
  • “Continue my onboarding from where I left off last session.”
  • “Walk me through setting my goals and initial tasks.”

Requirements

  • A Claude Code project using this starter kit's CLAUDE.md and rules
  • Pre-approved tools (allowed-tools): Read, Write, Edit, Bash, AskUserQuestion

Workflow steps

5 steps, taken from the step headings in SKILL.md.

  1. User Profile (5 min)
  2. 12 Favorite Problems (10 min)
  3. End Goal and Subgoals (5 min)
  4. Initial Tasks (5 min)
  5. AI Self-Knowledge (2 min)

What it can do on your machine

Read from SKILL.md and the folder at commit 546c72c. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves these tools, so the agent can use them without asking each time:

    • Read
    • Write
    • Edit
    • Bash
    • AskUserQuestion

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Shell commands in SKILL.md call:

    • git
    • python3
    • gh
    • brew

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md. Its commands use git and gh, which can reach the network depending on how they are called.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Personal Assistant Onboarding loads about 2.7k tokens when it runs. Until then it costs about 50 tokens; SKILL.md has 1,205 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~50
When it runs · the whole SKILL.md, loaded when a task matches
~2.7k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check: notes

The automated check noted patterns worth knowing about, such as sudo or a known installer.

  • NotePre-approves every shell command (allowed-tools: Bash)SKILL.md
    allowed-tools: Read, Write, Edit, Bash, AskUserQuestion

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from mp-web3/claude-starter-kit at commit 546c72c, republished under its MIT licence (© mp-web3). 1,205 words, ~2,661 tokens.

Download SKILL.mdSave it as .claude/skills/onboard/SKILL.md (or your agent's skills folder).
name
onboard
description
Guided first-session setup. Walks the user through building their personal AI assistant — profile, 12 problems, goals, subgoals, and initial tasks. Run this in the first session after setup.
allowed-tools
Read, Write, Edit, Bash, AskUserQuestion
argument-hint
[resume]

/onboard — Personal AI Assistant Setup

You are guiding a new user through setting up their personal AI assistant. This is their first session. Be conversational, curious, and patient. Don't rush through steps — each one matters.

Arguments: $ARGUMENTS

Before Starting

  1. Read ~/.claude/CLAUDE.md and all files in ~/.claude/rules/
  2. Read all files in knowledge/ (if any exist)
  3. Check if state/onboard-progress.yaml exists — if so, resume (see Resuming below)
Resuming

On ANY /onboard invocation:

  1. Check if state/onboard-progress.yaml exists
  2. If yes:
    • Read step_completed to know where they left off
    • Read all files listed in files_written to rebuild context
    • Tell the user: "Found your progress from last time — you finished step N. Picking up at step N+1."
    • Continue from step N+1
  3. If no: start from Step 1
  4. After completing the final step, delete onboard-progress.yaml and commit
Checkpoint Rule (Non-Negotiable)

After completing EACH step, immediately:

  1. Write/update state/onboard-progress.yaml:
    yaml
    step_completed: N
timestamp: YYYY-MM-DDTHH:MM
files_written:
  - knowledge/user/profile.md
  # ... list all files written so far
  2. Commit locally: git add -A && git commit -m "onboard: checkpoint after step N"

Do NOT batch commits to the end. Each step is a checkpoint. If the user disconnects, their progress is on disk.

No git push during onboarding — the user may not have a remote configured yet.

Greeting

Greet the user by name (from CLAUDE.md). Explain what you're about to do:

We're going to set up your AI assistant in 5 steps. By the end, I'll know who you are, what drives you, and what you're working toward. Every future session builds on what we create today.

This takes about 20-30 minutes. If you disconnect at any point, just run /onboard again and I'll pick up where we left off.

Step 1: User Profile (5 min)

Goal: Create knowledge/user/profile.md

Have a conversation. Ask one question at a time, wait for the answer. Don't dump a list of questions.

Start with:

Tell me about yourself. What do you do? What are you working on right now?

Then dig deeper based on their answers. You're trying to learn:

  • Career: what they do, where they work, tech stack, tools
  • Projects: what they're building or want to build
  • Communication style: observe how they write — short/long, formal/casual, emoji or not
  • Working preferences: IDE, OS, languages, frameworks

After 4-6 exchanges, summarize what you've learned and write knowledge/user/profile.md:

yaml
---
tags: [user, profile]
created: YYYY-MM-DD
last_reviewed: YYYY-MM-DD
---

Show them the file and ask: "Anything wrong or missing?"

Step 2: 12 Favorite Problems (10 min)

Goal: Create knowledge/problems/00-overview.md and individual problem files.

Introduce the concept:

Richard Feynman kept 12 problems constantly in mind. When he encountered a new idea, he tested it against each problem to see if it helped. We're going to define yours.

These aren't tasks or goals — they're the deep questions you keep coming back to. The things you think about in the shower. They rarely change (maybe once a year), and everything you do should connect to at least one of them.

Guide them through it:

  1. Ask: "What are the 3-4 things you think about most? Not tasks — the underlying questions or tensions."
  2. Listen. Reflect back what you hear. Help them articulate the deeper question behind their surface answer.
    • If they say "I want to build a SaaS" → "What's the deeper question? Is it about financial independence? About proving you can ship? About solving a specific problem?"
  3. Once you have 3-4, ask: "What else keeps you up at night? What do you wish you understood better?"
  4. Continue until you have 8-12. Some users will have 12 naturally. Some will have 6-8. Don't force it — 6 genuine problems beats 12 stretched ones.

Organize into layers:

LayerFocusExample
The CraftHow you build things"How do I ship faster without breaking things?"
The WorkWhat you're building"How do I find problems worth solving?"
The UnderstandingWhat you need to learn"How do I evaluate opportunities?"
The PersonWho you are becoming"How do I stay focused when everything feels urgent?"

Write knowledge/problems/00-overview.md with the Feynman quote, layers, cross-cutting notes, and "How to Use" section. Then create individual files (01-*.md through NN-*.md) for each problem with:

yaml
---
tags: [problems, feynman, LAYER_NAME]
created: YYYY-MM-DD
---

Each problem file should have: the problem statement, one-liner, and empty sections for "Insights Log" and "Open Sub-Questions" that will fill up over time.

Show the overview and ask: "Do these feel right? Any that don't belong, or any missing?"

Show full SKILL.md (549 more words)Show less

Step 3: End Goal and Subgoals (5 min)

Goal: Create knowledge/user/goals.md

Ask:

Looking at your 12 problems, what's the one big thing you're building toward right now? Not a specific project — the outcome. Where do you want to be in 6-12 months?

Help them articulate an end goal. Then break it into 3-5 subgoals:

What are the 3-5 milestones between here and that end goal? Each should be concrete enough that you'd know when it's done.

For each subgoal, capture:

  • What it is
  • "Done when" criteria
  • Which problems (from Step 2) it serves
  • Current status

Write knowledge/user/goals.md with:

  • End goal with "why"
  • Subgoals table (number, description, done-when, problems, status)
  • Three layers explanation (problems → subgoals → tasks)

Step 4: Initial Tasks (5 min)

Goal: Populate the task database and MEMORY.md with initial tasks.

First, insert the goals into the database:

bash
# Insert end goal
python3 -c "
import sys; sys.path.insert(0,'scripts')
from db import add_goal
add_goal('G0', '[END_GOAL_NAME]', 'end_goal', description='[END_GOAL_DESCRIPTION]')
"

# Insert each subgoal
python3 -c "
import sys; sys.path.insert(0,'scripts')
from db import add_goal
add_goal('S1', '[NAME]', 'subgoal', parent_id='G0', done_when='[CRITERIA]', problems='[1,2]', order_num=1)
"

Then ask:

What are the 3-5 most important things you need to do this week? What's blocking you or what are you procrastinating on?

For each task:

  • Connect it to a subgoal (flag if it doesn't connect to any)
  • Connect it to problems (flag if zero — "Is this drift?")
  • Assign priority: P1 (today/tomorrow), P2 (this week), P3 (this month)
  • Estimate size: S (< 1 session), M (2-3 sessions), L (4+ sessions)

Insert each task into the database:

bash
python3 -c "
import sys; sys.path.insert(0,'scripts')
from db import next_task_id, add_task
tid = next_task_id()
add_task(tid, '[TASK_NAME]', 'P1', problems='1,4', effort='S', subgoal='S1')
print(f'Created {tid}')
"

After all tasks are inserted, export to markdown:

bash
python3 scripts/db.py export

Update MEMORY.md with an Active Tasks table showing the top tasks.

Step 5: AI Self-Knowledge (2 min)

Goal: Create knowledge/self/identity.md

Create a brief self-knowledge file. This tells future Claude instances what they are in this user's system:

markdown
# Identity — What I Am

I am [USER_NAME]'s AI assistant, running on Claude Code. I have no memory between sessions —
everything I know persists through files on disk.

## What I Have
- Persistent knowledge in ~/claude-assistant/knowledge/
- Rules that shape my behavior in ~/.claude/rules/
- Session history in state/sessions/
- A cross-session index in MEMORY.md (first 200 lines auto-loaded)

## How I Improve
Corrections and preferences get written to files immediately.
Same correction twice → promote to a rule.
Every session ends with a session note and MEMORY.md update.

## Operating Principle
Be direct, honest, and useful. Don't hype, don't flatter, don't defer when I can act.
When I notice something the user should know, say it — don't wait to be asked.

Adapt the tone to match the user's communication style observed in Steps 1-4.

Finishing Up

  1. Final commit:

    bash
    git add -A && git commit -m "onboarding: complete"

  2. Delete progress file:

    bash
    rm state/onboard-progress.yaml && git add -A && git commit -m "onboard: cleanup progress file"

  3. Set up remote backup (optional but recommended):

    bash
    gh repo create claude-assistant --private --source . --push

    This creates a private GitHub repo and pushes everything in one command. Requires gh CLI (brew install gh && gh auth login).

    If they don't have gh, give the manual route:

    bash
    # Create the repo on GitHub first (github.com/new), then:
git remote add origin git@github.com:USERNAME/claude-assistant.git && git push -u origin main

  4. Update MEMORY.md with:

    • Active Tasks from Step 4
    • "Next Up" section pointing to the highest-priority task
    • File map of everything created

  5. Show a summary:

    Here's what we built:

    • Your profile: knowledge/user/profile.md
    • Your 12 problems: knowledge/problems/
    • Your goals: knowledge/user/goals.md
    • AI identity: knowledge/self/identity.md
    • [N] tasks in the database (view with /tasks)
    • Task backlog: state/backlog.md

    From now on, every session starts by reading these files. I'll remember who you are, what you're working toward, and what to do next.

    Skills available:

    • /tasks — manage your backlog (add tasks, review priorities, mark things done)
    • /plan — structured build workflow (explore, design, approve, implement, verify). Use this when building anything non-trivial.
    • /reflect — extract what I learned this session and write it to the right files. Run this periodically to close the learning loop.

  6. Create session note in state/sessions/YYYY-MM-DD-onboarding.md

If User Says "Pause"

At any point:

  1. Save all progress made so far (write partial files)
  2. Update state/onboard-progress.yaml with current step and files written
  3. Commit locally
  4. Tell them: "Progress saved. Next session, just run /onboard and I'll pick up where we left off."

© mp-web3, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in skills/onboard of mp-web3/claude-starter-kit.

Open the folder on GitHubat commit 546c72c

Compare with similar skills

Personal Assistant Onboarding next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Personal Assistant Onboarding compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Personal Assistant Onboarding this skillmp-web3/claude-starter-kit109—~2.7kAutomated safety check: NotesMIT
Using Superpowersfarm-fe/farm5.6k34 repos~1.4kAutomated safety check: PassMIT
Interview Meaddyosmani/agent-skills102k6 repos~3.8kAutomated safety check: PassMIT
Grillingbestofjs/bestofjs3.1k30 repos~464Automated safety check: PassMIT
Orca CLIstablyai/orca87k2 repos~593Automated safety check: PassMIT
Beads Task Memorygastownhall/beads28k—~1.2kAutomated safety check: PassMIT

Similar skills

  • Using Superpowers

    farm-fe/farm

    A skill your agent uses when starting any conversation - establishes how to find and use skills, requiring Skill tool invocation before ANY response including clarifying questions

    5.6k GitHub starsUsed in 34 repos~1.4k tokens
    Agent WorkflowsAuto-check passed

  • Interview Me

    addyosmani/agent-skills

    Asks one question at a time, each with a best guess attached, until the agent is about 95 percent sure what you really want, before any plan, spec or code.

    102k GitHub starsUsed in 6 repos~3.8k tokens
    Agent WorkflowsAuto-check passed

  • Grilling

    bestofjs/bestofjs

    Grill the user relentlessly about a plan, decision, or idea.

    3.1k GitHub starsUsed in 30 repos~464 tokens
    Agent WorkflowsAuto-check passed

  • Orca CLI

    stablyai/orca

    Operate Orca-managed worktrees, folder contexts, terminals, repos, automations, artifacts, skill sharing, worktree comments, and Orca's embedded browser…

    87k GitHub starsUsed in 2 repos~593 tokens
    Agent WorkflowsAuto-check passed

  • Beads Task Memory

    gastownhall/beads

    Tracks multi-session work with dependencies in the bd issue tracker so the agent can find ready tasks and recover its context after conversation compaction.

    28k GitHub stars~1.2k tokensUpdated today
    Agent WorkflowsAuto-check passed

  • Session History Search

    slopus/happy

    Searches past Claude Code, Codex and Cursor sessions and summarizes what was worked on, tried or decided, using extraction scripts instead of reading raw logs.

    24k GitHub stars~3.1k tokensUpdated today
    Agent WorkflowsAuto-check passed

More from mp-web3/claude-starter-kit

  • Create Skill

    mp-web3/claude-starter-kit

    Walks through building a new Claude Code skill: asks what it should do, scaffolds it with an init script, writes SKILL.md and LEARNINGS.md and validates the result.

    109 GitHub stars~985 tokensUpdated 6 mo ago
    Auto-check: notes

  • Claude Project Bootstrap

    mp-web3/claude-starter-kit

    Sets up a project for Claude Code by surveying it and writing a .claude folder with a short CLAUDE.md, a CODEBASE.md overview, module docs and optional skills and hooks.

    109 GitHub stars~1.9k tokensUpdated 6 mo ago
    Auto-check: notes

  • Plan and Implement Workflow

    mp-web3/claude-starter-kit

    Runs new features, scripts and structural changes through explore, design, approval, implementation and verification phases, with a tool search and an explicit plan sign-off.

    109 GitHub stars~913 tokensUpdated 6 mo ago
    Auto-check: notes

  • Session Reflection Learnings

    mp-web3/claude-starter-kit

    Analyzes recent sessions for corrections, preferences and implicit feedback, then routes the real learnings to the right knowledge or rules file.

    109 GitHub stars~1.8k tokensUpdated 6 mo ago
    Auto-check: notes

  • Task Backlog Manager

    mp-web3/claude-starter-kit

    Keeps a personal task backlog in SQLite behind a /tasks command that adds, reviews, completes and measures tasks against a list of long-term problems.

    109 GitHub stars~1.2k tokensUpdated 6 mo ago
    Auto-check: notes

Categories

Questions about Personal Assistant Onboarding

What does Personal Assistant Onboarding do?

Walks a new user through a five-step first session, profile, problems, goals and initial tasks, to turn the agent into a personal assistant that remembers them. The skill runs a guided, conversational setup meant for the very first session after installing a personal-assistant starter kit.md and any rule or knowledge files first, then checks a progress file to resume from the last completed step if the user disconnected partway through, telling them which step they left off on.

When should I use Personal Assistant Onboarding?

Personal Assistant Onboarding fits situations like: running the first session after installing a personal-assistant starter kit; resuming an onboarding conversation that was interrupted partway through; setting up a user profile, goals and initial tasks for a new assistant.

How do I install Personal Assistant Onboarding in Claude Code?

Run `npx skills add mp-web3/claude-starter-kit --skill onboard -a claude-code`. Or copy the skill folder (skills/onboard in mp-web3/claude-starter-kit) into .claude/skills/onboard in your project. Claude Code loads it when a task matches its description.

How do I install Personal Assistant Onboarding in Codex?

Run `npx skills add mp-web3/claude-starter-kit --skill onboard -a codex`. Or copy the skill folder (skills/onboard in mp-web3/claude-starter-kit) into .agents/skills/onboard in your project. Codex loads it when a task matches its description.

Can I use Personal Assistant Onboarding in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add mp-web3/claude-starter-kit --skill onboard -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/onboard, .gemini/skills/onboard, .github/skills/onboard and .opencode/skills/onboard in your project.

What does Personal Assistant Onboarding need to run?

Going by SKILL.md and its folder, Personal Assistant Onboarding needs the command-line tools its instructions call (git, python3, gh and brew). Our summary lists: A Claude Code project using this starter kit's CLAUDE.md and rules. Its frontmatter pre-approves these tools: Read, Write, Edit, Bash, AskUserQuestion.

Does Personal Assistant Onboarding access the network?

SKILL.md contains no URLs. Its commands use git and gh, which can reach the network depending on how they are called. This is read from the text; nothing was executed.

Is Personal Assistant Onboarding safe to install?

Our automated static check of SKILL.md found notes only (pre-approves every shell command (allowed-tools: bash)), nothing it rates as a warning. It is not a guarantee. Review the folder before installing.

What licence does Personal Assistant Onboarding use?

Personal Assistant Onboarding is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Personal Assistant Onboarding use?

About 2.7k tokens (SKILL.md is roughly 11k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Personal Assistant Onboarding?

Skills that share tags, products or a category with Personal Assistant Onboarding: Using Superpowers (farm-fe/farm, 5.6k stars), Interview Me (addyosmani/agent-skills, 102k stars), Grilling (bestofjs/bestofjs, 3.1k stars) and Orca CLI (stablyai/orca, 87k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Personal Assistant Onboarding?

mp-web3 (a GitHub user) maintains it in mp-web3/claude-starter-kit, which has 109 GitHub stars. The repository holds 6 skills in this directory. The repository was last updated on March 18, 2026.

Source: mp-web3/claude-starter-kit on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.