Clone Factory Playbook

Run the boarding orchestrator to create the workspace and kanban board.

Command: /boarding-orchestrator init {expert-name} --recon

What happens: Creates _workspaces/{expert-slug}/ with subdirectories, generates kanban.json with 24 cards, starts deep-research immediately.

If sources already exist: Use /boarding-orchestrator init {expert-name} (no --recon, 19 cards, skip to Phase 2)

Skill: deep-research (wraps Perplexity Sonar Deep Research API)

What it does: Runs 5 research queries against the expert's public presence. Produces a narrative intelligence report + structured data.

• Output: sources/deep-research-report.md, deep-research.json, deep-research-sources.json
• Expect: 100K-200K chars of raw research

Skill: expert-recon (orchestrates scraping + synthesis)

What it does: Scrapes the expert's website, social profiles, and key pages. Synthesizes with deep-research into a boarding readiness score.

• Output: sources/raw-sources/* (scraped content), {expert}-intelligence-report.html
• Boarding readiness score >= 7: sources sufficient, auto-proceed
• Score 4-6: partial sources, human should supplement
• Score < 4: need manual uploads

Skill: masterybook-sync

What it does: Uploads all sources to a MasteryBook notebook for team-accessible RAG Q&A. Optional — gracefully degrades if API unavailable.

• Output: sources/masterybook-sync-status.json, shareable notebook URL
• Runs in parallel with Step 5

Skill: expert-framework-creator

What it does: Reads ALL raw sources and discovers the expert's unconscious quality framework — the criteria they naturally apply but never codified. Uses 3-path convergence protocol.

• Path A: Framework-Forward (structure to content)
• Path B: Content-Back (content to structure)
• Path C: Anti-Pattern-First (failures to principles)
• Output: expert-framework.json with acronym(s), scoring rubrics, anti-patterns
• Expect: 1-3 framework acronyms (single, dual, or triple architecture)

Skill: soul-extractor

Output: soul.json — thinking loops, values hierarchy, governance rules, anti-patterns, background behaviors, canonical statements

Skill: voice-extractor

Output: voice.json — tone pairs, forbidden phrases, energy spectrum, signature elements, reading level

Skill: framework-extractor

Output: frameworks.json — tools/stacks, phases, taxonomies, routing pathways, cross-phase bridges

Skill: resource-extractor

Output: resources.json — books, podcasts, courses, tools, mentors, governance rules for recommendations

Skill: offer-extractor

Output: offers.json — offers, pricing tiers, conversion pathways, monetization governance (ZERO COERCION rules)

Skill: design-system-extractor

Output: Design system HTML — colors, typography, spacing, components. Optional if brand assets already documented.

Skill: gap-analyzer

What it does: Cross-references all extraction files, finds missing information, contradictions, and incomplete sections. Produces prioritized gaps.

• Output: gaps.json — each gap has severity (critical/important/nice-to-have), source file, specific field, and a question to resolve it
• 0 critical gaps = proceed to compilation
• Any critical gap = STOP and resolve before continuing

Skill: rubric-builder

What it does: Generates 6-dimension scoring rubrics from extractions + expert framework. These rubrics score every downstream artifact.

• Output: rubrics/rubrics.json
• 6 dimensions: voice fidelity, framework accuracy, governance compliance, completeness, expert quality (FORGE or equivalent), universal quality (GOLDEN+SHARP)
• Requires expert-framework.json — if missing, warns and builds 5-dimension version

Skill: clone-compiler

What it does: Reads all 7 extraction files + expert-framework.json and compiles them into a complete AI coaching OS.

• artifacts/system-prompt.md — 14-section system prompt (30-50KB)
• artifacts/knowledge-file-part1.md — active phases with full stack specs
• artifacts/knowledge-file-part2.md — coming-soon phases + resources
• artifacts/tool-configs.json — machine-readable tool registry

Skill: lead-magnet-builder

Output: artifacts/lead-magnet.html — standalone assessment page with questions, scoring, result types, email capture

Skill: onboarding-builder

Output: artifacts/onboarding-flow.html — Netflix-style card-based onboarding with discovery questions and tool routing

Skill: clone-tester

What it does: Runs the compiled clone through 25-50 simulated scenarios across 5 categories. Scores against 8 dimensions. Produces deployment decision.

• test-results/test-results.json — full scored results
• test-results/audit-sheet.md — human-readable audit document
• test-results/simulation-log.json — all test scenarios with responses

What: Generate an interactive audit form from the test results and expert review questions. Publish to NowPage for the expert and team to fill out.

Generic Audit Form Template (10 Screens)

When audit form is submitted:

• Read the audit report from NowPage
• Apply any corrections to extraction files (voice adjustments, terminology changes, pricing confirmations)
• If significant changes: re-run clone-compiler and clone-tester
• If minor changes: patch system-prompt.md directly
• Publish updated audit page with response incorporated

Publish to NowPage:

• /{brand}-pipeline-run-{N} — Pipeline run detail (scores, heatmap)
• /{brand}-build-registry — Chapter table for this expert's builds
• Update /factory-build-registry with new expert entry
• Update /boarding-pack with deployment status