pos: Skill for Claude Code

/vault-scaffolder — Personalize the Personal OS (in-place)

Takes a completed svoboda profile.yaml (v3.2 schema) and personalizes the Personal OS in the current workspace. Overwrites the blank onboarding CLAUDE.md, writes context/identity.md, context/goals.md, the self-entity, and the operational layer, and appends profile-specific anti-patterns to the shipped corpus.

Assumes the engine is already present. Skills (.claude/skills/), templates (_templates/), context/scoring-gate.md, AND the shipped engine corpus — context/anti-patterns.md, anti-patterns-index.md, context/learned.md, learned-index.md, rules/*.md — all ship with the starter repo. The scaffolder personalizes content; it does NOT re-scaffold skills, templates, or the AP/L corpus. It NEVER regenerates anti-patterns.md or learned.md — it appends to them.

Reads the v3.2 nested schema directly (see contract below). It does not require the profiler to flatten anything.

Invocation

/vault-scaffolder {subject_id}

Input

1. Required: {subject_id} argument
2. Read memory/svoboda/{subject_id}/profile.yaml
3. If missing → error: "Run /svoboda-profiler {subject_id} first"

profile.yaml — schema contract (v3.2, must match svoboda-profiler)

This is the SAME canonical schema defined in svoboda-profiler/SKILL.md Phase 4. Read these exact keys. Nesting matters.

subject_id: str                                   # kebab-case, == folder name
profile_version: str                              # top-level (e.g. "2.0")
synthesized: str                                  # YYYY-MM-DD

identity:                                         # NESTED map
  name: str
  age: int
  role: str
  languages: [str, ...]                           # languages[0] = primary
  timezone: str                                   # IANA tz

svoboda_scores:                                   # NESTED map, 1-10 each (OBSERVED)
  samorazvitie: int
  vitalnost: int
  okruzhenie: int
  bogatstvo: int
  otdyh: int
  delo: int
  aktivy: int

domains_active: [str, ...]                        # FLAT list
north_star: str                                   # 1-2 sentences, verbatim into goals.md
style: direct|explanatory|mixed
cognitive_style: scanning|deep-dive|burst|gradual
triggers: [str, ...]                              # FLAT list, 3-5 items
blind_spots: [str, ...]                           # FLAT list
predictive_model: [{situation: str, behavior: str}, ...]   # list of rows

tags:                                             # NESTED map (NOT a flat list)
  core_confirmed: [str, ...]
  refined: [str, ...]
  retired_from_baseline: [str, ...]

synthesis_constraint: str                         # optional free-text guardrail

# Optional / advisory (read if present, otherwise skip — do NOT fail):
score_divergence: [...]
granularity_coverage: {...}
cadence: {chronotype, reflection}
# plus v3.2 deep-layer fields (attachment, attribution, ...) — NOT consumed here.