agentleFS
Pearl1531Pearl1531/Archetype-talk/.claude/skills/welcome/SKILL.md

welcome

Skill2 starsChanged 50 days agoMITView on GitHub

First-run onboarding — greet a new user, explain what Archetype Talk is and whose it is, list capabilities, and route to setup or the demo persona. Runs via the SessionStart hook on fresh installs or on /welcome.

  • Reads credentials

8 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

6 rules. 0 are common, 5 are distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 7 edited versions we read keep these.

  • 5

    Distinctive

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Do this before you write a word: Read project_name: in Product Context.md and check .claude/cache/graph-index.md for its project: and …

    line 14
  2. Distinctive:

    Keep it warm and short. Do not dump everything at once.

    line 32
  3. Distinctive:

    Read README.md for the owner/author line and license.

    line 36
  4. Distinctive:

    Ask the user (one clear question):

    line 55
  5. Distinctive:

    Only what the project genuinely needs to continue:

    line 60
  6. Distinctive:

    When onboarding is done, create the marker so it doesn't repeat:

    line 73

“Common” is measured: kept by at least 6 in 10 of the 7 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

welcome — First-run onboarding

Greet a new user right after they install/clone this repo, explain what it is, describe what you can do, and route them to either configuration or hands-on work. Runs automatically on first session (via the SessionStart hook, when .claude/onboarded.local is missing) or on demand with /welcome.

Adapt to the user's language. The repo ships in English by default, but detect the user's language from how they write and respond in it. Translate section labels and your prose; never translate verbatim participant quotes (they are data).

Step 0 — find out whether this repo is already somebody's project

Do this before you write a word. Read project_name: in Product Context.md and check .claude/cache/graph-index.md for its project: and demo_entities: lines. A repo that already carries a project is not a fresh install, and the worst opening you can give is a tour of the demo to someone whose own research is sitting right there.

  • A project is named and real (non-demo: true) entities exist → say what the project is in your first sentence, from Product Context.md. Offer to pick up the work; keep the capability list short and skip the demo entirely unless they ask. Never present a demo persona alongside their own.
  • Nothing but the demo → the flow below, as written.

The rest of this skill assumes the second case. In the first, use the parts that apply and drop the rest — an onboarding that ignores what is already in the folder is worse than no onboarding.

The onboarding flow — four steps, one at a time

Keep it warm and short. Do not dump everything at once.

Step 1 — What this is & whose it is

  • This is Archetype Talk — a UX-research knowledge graph you can talk to. It turns real customer research into intention-based personas grounded in traceable evidence.
  • Read README.md for the owner/author line and license. State whose repo it is (author from README / LICENSE) so the user knows they're standing in someone's method, not a blank tool.
  • One-sentence shape: Evidence → Signal → (Persona ↔ Archetype) → Idea, with optional Competitors/ context.

Step 2 — What you can do here

Briefly list your capabilities (map to the skills):

  • Talk to a persona grounded strictly in the research (persona-talk) — incl. a ready DEMO persona: Personas/Emma.md.
  • Query a persona for likely reactions with confidence levels (persona-query).
  • Build / deepen personas from research (persona-workshop, ai-persona).
  • Start from absolute zero — no transcripts, no reports at all: /cold-start interviews the founder about the startup (value proposition, target group, pitch deck) and turns their beliefs into testable L1 assumptions + a research plan. Explicitly invoked only.
  • Turn transcripts into Signals & Evidence (extract-findings).
  • Connect research sources over MCP — Dovetail (dovetail-sync); voice memos & call recordings, Notion/Confluence, usability tools (source-sync); product analytics: Mixpanel / GA4 / Amplitude (analytics-sync). All read-only, with provenance.
  • Keep the graph healthy — /graph-lint (dead links, orphans, stale data), /contradictions, /backlog.
  • Close the research loop — /interview-guide turns backlog gaps into a discussion guide for real interviews.
  • PM toolkit — /feature-panel (one feature × all personas, cited), /prioritize (RICE with Confidence derived from Levels), /prd (fully-cited Why), /opportunity-tree (OST from Correlations).
  • Deliver outward — /journey-map (cited journey maps), /export (persona one-pagers, Ideas as tickets, stakeholder packs).
  • Give a persona an ElevenLabs voice so /persona-talk replies can be heard, not just read (persona-voice) — needs the user's own ElevenLabs API key.
  • Everything is graded by grounding Levels (L1–L5, one scale with persona maturity) and freshness (3-month rule), and every claim links back to a real source.

Step 3 — Ask which way to go

Ask the user (one clear question):

"Would you like to (a) go through a short setup so the repo fits your project, or (b) jump straight in — talk to the DEMO persona or add your own research?" Mention explicitly that they can try the DEMO persona right now with /persona-talk Personas/Emma.md, no setup needed.

Step 4 — If they choose setup, ask the essentials (one at a time)

Only what the project genuinely needs to continue:

  1. Product under study — what product/service are the personas about? (updates Product Context.md)
  2. Research source — do they have transcripts to drop in Inbox/, or a Dovetail workspace to connect via /dovetail-sync? (or neither yet → offer /cold-start for a zero-data founding brief, or /persona-workshop to sketch an archetype)
  3. Primary segment(s) — who are they researching first? (shapes the first persona/archetype) 3a. Competitor compare categories (only if they'll track competitors) — which 3–6 axes genuinely decide wins and losses in their market (pricing model, the contested capability, ownership, support, compliance…)? Write them to compare_categories: [...] in Product Context.md frontmatter — the app's Compare view shows exactly these. Steer them away from vanity metrics (funding, headcount, follower counts): if it doesn't change what their users choose, it's not a category.
  4. Keep, separate, or clear the example data? — the Spotify example set is illustrative; every file carries demo: true in frontmatter (a Demo badge in the app, ⚠️ flagged by persona-talk). Three options via /demo-data: keep in place (badges mark it), separate it under Demo/ (recommended once real data starts flowing), or delete it. No pressure to decide now — persona-talk will ask once after their first demo session anyway. Real content they add should simply omit the demo: field.
  5. Language — confirm the language they want to work in (you adapt; repo stays English by default).
  6. Voice (optional) — want personas to have an ElevenLabs voice? Needs their own API key in .env (ELEVENLABS_API_KEY); if they don't have one yet, skip this for now — it's revisitable anytime via persona-voice, not a one-time choice.

Then propose the concrete next action (connect Dovetail / process Inbox / start a workshop) and, if useful, remind them the DEMO persona is available.

Finish

When onboarding is done, create the marker so it doesn't repeat:

touch .claude/onboarded.local

(If the user only ran /welcome manually and was already set up, don't recreate the marker logic — just help.)

Rules

  • Never fabricate the author/product — read README.md / LICENSE for the owner; if the product isn't set yet, say so and offer to set it in Product Context.md.
  • Don't auto-run heavy work. Onboarding never bulk-analyzes files on its own — it points to /extract-findings and asks first.
  • One question at a time in Step 4; summarize what you heard before moving on.
  • Never offer the demo to a repo that already has a project. Step 0 decides this, and demo: true is the only thing separating the example set from the user's own files (hard rule 9 in CLAUDE.md) — so an unfiltered list of personas is always wrong here.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.