Kp Content Authoring

Parallel Instruction Patterns (NFR-06)

The following instruction sections are identical across all 35 packs and should be cached:

1. Pack file structure template -- the 6-file directory layout is the same for every pack
2. Vision document skeleton -- section headings (Vision, Problem Statement, Core Concepts, Skill Tree, Modules, Assessment Framework, Parent Guidance, Resources) are reused
3. Modules YAML schema -- PackModuleSchema structure with time_estimates, prerequisite_modules, activities sub-objects
4. Activities JSON schema -- PackActivitySchema with grade_range, duration_minutes, materials, learning_objectives
5. Assessment rubric template -- 4-level rubric (Beginning, Developing, Proficient, Advanced) with formative/summative sections
6. Resource catalog template -- categorized sections (Young Learners, Older Learners, Parents, Deeper Study)
7. .skillmeta boilerplate -- common fields (version, status, copyright, gsd_integration, accessibility stubs, translation stubs)
8. Grade level bands -- Foundation (PreK-K), Elementary (1-5), Middle School (6-8), High School (9-12), College (13-16)

These 8 patterns are loaded once and reused across all pack generation sessions, reducing per-pack token overhead by ~60%.

Key Modules

Usage Examples

Generate a pack content suite:

For each pack (e.g., MATH-101):
1. Create src/knowledge/packs/math-101/ directory
2. Write vision.md using pack-specific content + shared template sections
3. Write modules.yaml matching ModulesFileSchema
4. Write activities.json as PackActivity[] array
5. Write assessment.md with 4-level rubric
6. Write resources.md with categorized links
7. Write .skillmeta matching KnowledgePackSchema

Apply parallel instruction patterns:

Before generating any pack, load:
- Pack file structure template (shared)
- Vision document skeleton (shared)
- Modules YAML schema (shared)
- Activities JSON schema (shared)
- Assessment rubric template (shared)
- Resource catalog template (shared)
- .skillmeta boilerplate (shared)
- Grade level bands (shared)
Then inject pack-specific: subject content, domain activities, subject-specific resources

Dependencies

• Pack runtime API (src/knowledge/) for schema validation
• Domain-specific skill (core, applied, or specialized) for subject matter knowledge
• Pack template structure from delivery package

Token Budget Rationale

2.0% budget reflects the comprehensive authoring methodology covering 6 file types, 8 parallel instruction patterns, and the pack generation workflow. This is the largest knowledge pack skill because it provides the reusable templates that reduce per-pack generation cost. Shared across 3 agents but loaded once per session.