Structured Data
Capability Metadata
| Field | Value |
|---|---|
| ID | CAP-103 |
| Version | v1.27 |
| Domain | Specialist Domains |
| Group | Knowledge Packs |
| Status | Documented |
| Related | CAP-057, CAP-101, CAP-104 |
| Output | GSD skill file in .claude/commands/ |
| Demotion | Knowledge decay triggers re-evaluation |
Configuration Reference
Promotion threshold (from CAP-104 frontmatter):
exercises_completed — minimum exercises done
mastery_score — minimum score (0-1)
observation_count — minimum pack activations
days_active — minimum learning duration
Demotion triggers: knowledge decay detected (no domain sessions for 60+ days)
interface PromotionCriteria {
exercises_completed: number; // min exercises done
mastery_score: number; // min score (0-1)
observation_count: number; // min pack activations
days_active: number; // min learning duration
}
// When criteria met, generate skill from knowledge pack
async function promoteToSkill(
pack: KnowledgePack,
learner: LearnerState
): Promise {
const patterns = pack.getLearnedPatterns(learner);
return skillGenerator.fromKnowledgePack(pack, patterns);
} Technical Specifications
Connects to: core promotion pipeline (CAP-057) for skill generation and pipeline registration
Threshold settings: defined in knowledge pack frontmatter (CAP-104 promotion_threshold field)
Promoted skill: activates from observation alone in future sessions — no knowledge pack re-engagement required
Skill template: generated from knowledge pack's skills section + learner's demonstrated patterns
Overview
Knowledge Promotion closes the loop: domain expertise accumulated through knowledge pack exercises becomes permanent GSD skills that activate automatically in future sessions. When a learner demonstrates mastery of knowledge pack content — completing exercises above threshold, maintaining high concept scores across multiple sessions — the pack generates a GSD skill capturing the learned patterns.
Knowledge Promotion connects the knowledge pack subsystem to the core skill pipeline (CAP-057). The promotion criteria are defined in the knowledge pack's YAML frontmatter (CAP-104 format): minimum exercises completed, minimum mastery score, minimum observation count, and minimum days active. When all four thresholds are satisfied, Knowledge Promotion calls the skill generator to create a new GSD skill from the knowledge pack's skills section and the learner's demonstrated patterns.
The promoted skill is permanent and session-independent: it activates from observation alone whenever the domain is detected in a future session, without requiring re-engagement with the knowledge pack. This is the payoff of the knowledge pack learning loop — domain expertise that was once trapped in an interactive learning experience becomes autonomous, always-available context.
History
Before promotion, knowledge pack learning was session-local — users would learn AGC patterns in a knowledge pack session, then need to re-engage the pack in every subsequent session. The knowledge didn't transfer to the broader skill system. Each session started fresh from the same novice entry point, regardless of how much the learner had already demonstrated.
Knowledge Promotion was designed specifically to close this gap by connecting to the existing core promotion pipeline (CAP-057). Rather than building a separate promotion mechanism for knowledge packs, the design reuses the same promotion pipeline that promotes workflow patterns into skills — knowledge packs simply generate a different input to that pipeline. The four-criteria threshold (exercises, score, observations, days_active) was calibrated to prevent premature promotion: a learner who completes exercises quickly but only in a single intensive session shouldn't generate a permanent skill. The days_active requirement ensures skills represent genuine long-term learning, not weekend cramming.
Key Concepts
Imagine you spend a month learning AGC assembly programming through a knowledge pack — working through exercises, making mistakes, getting hints, and gradually mastering the instruction set. Knowledge Promotion is the graduation ceremony: when you've demonstrated enough mastery, the system creates a permanent "AGC Expert" skill that follows you into every future project.
From that point forward, whenever you're working on an Apollo-related project, your promoted skill activates automatically. You don't need to open the knowledge pack again. The knowledge pack was the school; the promoted skill is the diploma — proof of learning that travels with you.
Usage Examples
Example 1: Checking promotion eligibility
const eligibility = await promotionEvaluator.check({
learnerId: 'user-123',
pack: 'apollo-agc-basics'
});
console.log(eligibility);
// {
// eligible: true,
// exercises_completed: 8 / 5 required,
// mastery_score: 0.87 / 0.75 required,
// observation_count: 12 / 3 required,
// days_active: 14 / 7 required
// }Example 2: Generated skill after promotion
# .claude/commands/apollo-agc-basics-promoted.md
# Auto-generated by Knowledge Promotion (CAP-103)
# Source: knowledge pack apollo-agc-basics v1.0
# Learner: user-123, promoted: 2026-02-23
---
name: apollo-agc-basics-promoted
description: >
Apollo AGC assembly programming expertise.
Activates automatically when AGC code is detected.
activation:
pattern: "agc|dsky|verb noun|tc |ccs "
confidence: 0.7
---
# Applies knowledge from apollo-agc-basics knowledge pack
# Mastery: practitioner (score 0.87 across 14 sessions)