# Setup Wizard > Run the Cowork workspace onboarding wizard — personalized setup for your goal (study, research, writing, PM, creative, business, personal-assistant) - **Type:** Skill - **Install:** `agentstack add skill-jmlozano1990-cowork-starter-kit-setup-wizard` - **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior - **Seller:** [jmlozano1990](https://agentstack.voostack.com/s/jmlozano1990) - **Installs:** 0 - **Category:** [AI & ML](https://agentstack.voostack.com/c/ai-and-ml) - **Latest version:** 0.1.0 - **License:** MIT - **Upstream author:** [jmlozano1990](https://github.com/jmlozano1990) - **Source:** https://github.com/jmlozano1990/Cowork-Starter-Kit/tree/main/.claude/skills/setup-wizard - **Website:** https://github.com/jmlozano1990/Cowork-Starter-Kit/releases/latest ## Install ```sh agentstack add skill-jmlozano1990-cowork-starter-kit-setup-wizard ``` Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent. ## About ## Setup Wizard This skill runs the Cowork workspace onboarding interview. **WIZARD.md is the single script source — this file only routes into it.** Locate it at `WIZARD.md` (fresh kit) or `_setup-kit/WIZARD.md` (after the Step 7 handover archived the installer); the same rule applies to `skills/`, `selection-presets.md`, and every other kit path the script references. Never re-ask a field already answered in this session or recorded in `cowork-profile.md`. **Resume guard:** if `cowork-profile.md` exists with `Status: in-progress`, do NOT reset — resume per WIZARD.md Fallback (recap goal + bundle, continue from the first unanswered field). **Reset guard:** if `cowork-profile.md` exists with real completed content, first offer the friendly menu: "You already have a workspace set up. 1) Keep it as-is 2) Add or remove skills 3) Start fresh from a new goal". Only option 3 triggers the reset confirmation: "This will reset your profile and re-run onboarding. Your past sessions are unaffected. Confirm? (Yes / No)" — proceed only on Yes; on No, return to the menu. --- ## Interview Run WIZARD.md from Q1. Use AskUserQuestion buttons when available; otherwise numbered lists, with `**Your answer:**` on its own line. ### Q1 — Goal > "Welcome! What do you need help with? Describe your goal in your own words — or type 'not sure' for suggestions." Quoted verbatim from WIZARD.md's canonical Q1 opener (Single-Source Rule) — do not paraphrase it here. If the user is uncertain, defer to WIZARD.md's Uncertainty Fallback (Q1) rather than re-presenting a preset menu in this file. Route per WIZARD.md Q1 (Path A/B/C, stemmed signals, judgment tie-break). Every path produces a **draft** the user shapes — the 7 presets are starting suggestions, not fixed selections, and a custom Path C draft is equally first-class. ### Everything after Q1 Follow WIZARD.md in order: F4 bundle customization → profile-stub checkpoint → fast-track offer (exactly once, there) → Q2 (name + role + deadlines, one turn) → safety notice → optional Q3 voice turn → After-Q2 generation steps → Step 7 handover (generate the personalized workspace CLAUDE.md, archive the installer to `_setup-kit/`, offer working folders). The interview is 3 core turns plus the optional voice turn; output format is defaulted from the preset, tools/connectors are asked at point-of-need, and all skills install by copying from the local `skills/` pool — no internet needed (WIZARD.md Network & Offline Rule). --- ## After the interview `cowork-profile.md` follows the WIZARD.md Step 1 template — the only profile schema. Close with the task-first message (WIZARD.md Closing), using the matching first-task invitation: - Study: "Your [Subject] study space is ready. Paste something you're studying and I'll turn it into flashcards or notes." - Research: "Your [Domain] research workspace is ready. Share 2–3 sources and I'll synthesize them." - Writing: "Your writing space is ready. Paste a paragraph you've written and I'll match your voice, or let's outline something new." - Project Management: "Your [Role] workspace is ready. Tell me where a project stands and I'll draft your first status update." - Creative: "Your creative workspace is ready. Give me a half-formed idea and I'll help you develop it." - Business/Admin: "Your workspace is ready. Forward me the gist of an email you need to write and I'll draft it." - Personal Assistant: "Your workspace is ready. Tell me what's on your plate this week and I'll organize it." - Custom (Path C): "[Goal] workspace is ready. [One-sentence invitation using an installed skill on their actual goal.]" **Skill verification (once, not per skill):** confirm each installed file exists at `.claude/skills//SKILL.md` and say so in one line. Cowork auto-discovers these files; `skills-as-prompts.md` is the fallback for surfaces that don't. Always ask for explicit confirmation before deleting, moving, or overwriting any file or folder. ## Source & license This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code. - **Author:** [jmlozano1990](https://github.com/jmlozano1990) - **Source:** [jmlozano1990/Cowork-Starter-Kit](https://github.com/jmlozano1990/Cowork-Starter-Kit) - **License:** MIT - **Homepage:** https://github.com/jmlozano1990/Cowork-Starter-Kit/releases/latest Install and usage instructions live in the source repository linked above. ## Pricing - **Free** — Free ## Security capabilities Automated source analysis of v0.1.0 — what this tool can access: - **Network access:** no - **Filesystem access:** no - **Shell / process execution:** no - **Environment & secrets:** no - **Dynamic code execution:** no *"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.* ## Versions - **0.1.0** — security scan: passed — Imported from the upstream source. ## Links - Listing page: https://agentstack.voostack.com/l/skill-jmlozano1990-cowork-starter-kit-setup-wizard - Seller: https://agentstack.voostack.com/s/jmlozano1990 - Browse the marketplace: https://agentstack.voostack.com/browse --- Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.