career-hub-init

career-hub-init — Scaffold a Career Hub

Scaffold an empty, compliant career hub at the current working directory.

When Claude Invokes This

The user wants to start a fresh career hub. Typical prompts:

• "Create a career hub here"
• "Scaffold a career-facts repo"
• "Start a new personal knowledge base for my CV / LinkedIn content"

See ../references/hub-detection.md for when to refuse.

Hub Structure

The hub produced by this skill has this mandated structure:

<hub>/
├── PRINCIPLES.md
├── README.md
├── a.foundations/
│   ├── identity.md
│   ├── positioning.md
│   ├── biographical-facts.md
│   └── operating-notes.md
├── b.history/
├── c.projects/
├── d.capabilities/
├── e.artefacts/
│   ├── README.md
│   └── index.md
├── f.pipeline/
│   ├── target-roles.md
│   ├── unwritten-case-studies.md
│   ├── open-threads.md
│   └── application-log.md
├── g.backlog/
│   └── README.md
└── docs/superpowers/
    ├── specs/
    └── plans/

Principles (seeded into PRINCIPLES.md)

1. Category precedence. Lower-letter category wins; others cross-link.
2. Visibility is explicit. Every file declares public | private | confidential.
3. Confidential sibling pattern. .private.md sibling for candid retrospection.
4. Status markers. active | draft | archived. No deprecated.
5. Templates for all docs.
6. Stubs, not blanks. init seeds stubs with <!-- TODO --> markers.
7. Canonical-here vs external. e.artefacts/index.md tracks both with a migration rule.
8. Hub-first. Source of truth.
9. Draft-from-known-content-ask-only-gap-questions.
10. Every claim verifiable.

Behaviour

Step 1 — Confirm

Check that cwd:

• Does not already contain a hub (no PRINCIPLES.md, no full category directories — see hub-detection.md).
• Is empty or contains only .gitignore, README.md, or similar scaffolding remnants.

Ask the user:

Create a career hub at <cwd>?

This will create approximately 40 files:
- PRINCIPLES.md (governance), README.md (index)
- 7 category directories (a.foundations through g.backlog)
- a.foundations: 4 seeded files (identity, positioning, biographical-facts, operating-notes)
- b.history, c.projects, g.backlog: category READMEs only (populate via `build`)
- d.capabilities: 6 skeleton capability cards
- e.artefacts: README + index.md
- f.pipeline: 4 working docs (target-roles, unwritten-case-studies, open-threads, application-log)
- docs/superpowers/{specs,plans}/

Proceed? (y/n)

On explicit y, proceed. On n or ambiguous response, stop.

Step 2 — Create Directory Structure

Use Bash:

mkdir -p <cwd>/{a.foundations,b.history,c.projects,d.capabilities,e.artefacts,f.pipeline,g.backlog,docs/superpowers/specs,docs/superpowers/plans}

Step 3 — Seed Files

Use Write to create each file per the templates in ../references/templates.md. File list and contents:

Root level:

• PRINCIPLES.md — full governance document (see below for content skeleton).
• README.md — cross-category index table.
• .gitignore — standard ignores.

a.foundations/ (4 files):

• identity.md, positioning.md, biographical-facts.md, operating-notes.md — all from templates.md templates, with <!-- TODO --> markers seeded for the user to fill in.

b.history/, c.projects/, g.backlog/ — empty at init except:

• b.history/README.md — one-line pointer: "Populate via build roles."
• c.projects/README.md — one-line pointer: "Populate via build projects."
• g.backlog/README.md — from the g.backlog/README.md template.

d.capabilities/ (6 skeleton capability cards):

• languages.md, data-stack.md, ml-statistics.md, domain-expertise.md, infra-platforms.md, credentials.md — each from the d.capabilities template, with capability frontmatter field set to the file's topic and body as <!-- TODO --> markers.

e.artefacts/ (2 files):

• README.md — directory rationale template.
• index.md — artefact index template.

f.pipeline/ (4 files):

• target-roles.md, unwritten-case-studies.md, open-threads.md, application-log.md — all from their respective f.pipeline templates.

Step 4 — PRINCIPLES.md skeleton

The PRINCIPLES.md created at init must include at minimum:

---
visibility: private
audience: both
status: active
last_updated: <today>
---

# PRINCIPLES — Governance for this Career Hub

## Quick Reference

<table summarising: category precedence, visibility filter for publish, .private.md convention>

## Why This Directory Exists

<Hub-first philosophy, 2-3 sentences>

## Visibility Taxonomy

<copy from ../references/taxonomy.md axis 2>

## Confidential Sibling Files (`.private.md`)

<copy from taxonomy.md>

## Audience Taxonomy

<copy from taxonomy.md axis 3>

## Status Markers

<copy from taxonomy.md axis 4>

## Category Precedence

<copy from taxonomy.md>

## Templates

<pointer to templates.md in skill package; note that each category has a mandated template>

## Confidential Sibling Pattern

<how to create and use .private.md siblings>

## Cross-Linking Convention

<relative links between categories>

## Prior Art

<acknowledge olegvg/resume-tailor-plugin, nyinyinyanlin/profile-vault-obsidian, erichowens/career-biographer, jacksenechal/resume, demansou/resume-builder>

Step 5 — README.md skeleton

The README.md created at init must include:

# <hub-name> — Canonical Career Hub

> See [PRINCIPLES.md](PRINCIPLES.md) for governance.

Canonical source of career facts for <user>. Downstream CVs, GitHub profiles, LinkedIn content, and published case studies render from here.

## a. Foundations

| Document | Visibility | Summary |
|----------|------------|---------|
| [identity](a.foundations/identity.md) | private | Name, location, contact, public links |
| [positioning](a.foundations/positioning.md) | private | Alternative professional narratives by audience |
| [biographical-facts](a.foundations/biographical-facts.md) | private | Verifiable claims with verification methods |
| [operating-notes](a.foundations/operating-notes.md) | private | Cross-surface consistency rules |

## b. History

*Populate via `build roles`.*

## c. Projects

*Populate via `build projects`.*

## d. Capabilities

| Document | Visibility | Summary |
|----------|------------|---------|
| [languages](d.capabilities/languages.md) | public | Programming and data languages |
| [data-stack](d.capabilities/data-stack.md) | public | Modern data tooling |
| [ml-statistics](d.capabilities/ml-statistics.md) | public | ML and statistical methods |
| [domain-expertise](d.capabilities/domain-expertise.md) | public | Domain knowledge (credit risk, etc.) |
| [infra-platforms](d.capabilities/infra-platforms.md) | public | Infrastructure and platforms |
| [credentials](d.capabilities/credentials.md) | public | Education, awards, public artefacts |

## e. Artefacts

| Document | Purpose |
|----------|---------|
| [README](e.artefacts/README.md) | Why this directory is pointers-only |
| [index](e.artefacts/index.md) | Pointer table to rendered outputs |

## f. Pipeline

| Document | Visibility | Summary |
|----------|------------|---------|
| [target-roles](f.pipeline/target-roles.md) | private | Current target roles and application status |
| [unwritten-case-studies](f.pipeline/unwritten-case-studies.md) | private | Prioritised case-study backlog |
| [open-threads](f.pipeline/open-threads.md) | private | Loose ideas and half-started writeups |
| [application-log](f.pipeline/application-log.md) | private | Chronological application log |

## g. Backlog

| Document | Visibility | Summary |
|----------|------------|---------|
| [README](g.backlog/README.md) | private | Index of open threads deferred from interviews |

Step 6 — git init and first commit

cd <cwd>
git init
git add -A
git commit -m "chore: initial scaffold of career hub per career-hub-init v1"

If the user already has cwd under git, skip git init and add-commit on the current branch.

Step 7 — Print next-step suggestion

Career hub scaffolded at <cwd>.

Next steps:
- `career-hub-build foundations` — fill identity, positioning, biographical facts, operating notes.
- `career-hub-build roles` — start populating role cards.
- `career-hub-build projects` — start populating project case studies.
- `career-hub-build capabilities` — enrich the 6 seed capability cards.

Or `career-hub-build` with no argument to get a menu.

Prior Art

This skill draws on and differentiates from:

• olegvg/resume-tailor-plugin — single-file master profile, per-role visibility tags.
• nyinyinyanlin/profile-vault-obsidian — folder-per-category Obsidian vault.
• erichowens/career-biographer + cv-creator — interview-plus-render split, empathetic interview style.
• jacksenechal/resume — markdown resume with branch-based tailoring.
• demansou/resume-builder — evidence-mining, confidentiality-by-design.

Differentiators: category taxonomy with precedence; .private.md siblings; gap-question-only interview posture; e.artefacts/ with migration rule; g.backlog/ category; hub-first philosophy.

Refuses

• cwd already contains a hub. Direct the user to build.
• cwd is non-empty with content not matching init-compatible scaffolding.

Does Not

• Populate role / project content beyond stubs (that is build's job).
• Assume a remote git host — the user decides whether to push.
• Publish anything.