knowledge-architect

name	knowledge-architect
description	Consolidate digital signage, SignSync, Media OS, SaaS signage, restaurant POS signage, CCTV/media nexus, Tauri desktop, and research documentation into a domain-first, source-traced Media OS knowledge base.

Media OS Knowledge Architect

Use this skill to process fragmented signage documentation into the canonical media-os-knowledge/ structure. Raw project folders are evidence and must remain unchanged.

Product Center

Media OS is canonical. SignSync, SaaS signage, restaurant/POS signage, CCTV, IPTV/radio, AI, and fleet management are modules or product layers around a local-first Media OS runtime.

Output Structure

Maintain this structure:

media-os-knowledge/
  agent-index.md
  source-manifest.md
  00-index.md
  01-source-inventory.md
  02-domain-map.md
  03-work-spec.md
  04-feature-matrix.md
  05-workflows.md
  06-architecture-decisions.md
  07-improvement-backlog.md
  domains/
    player-runtime/
    layout-composition/
    content-assets/
    media-sources/
    restaurant-pos/
    device-fleet/
    sync-offline/
    orchestration/
    security-licensing/
    ai-optimization/
    research-strategy/
  extraction-cards/

Each domain folder must contain:

• README.md
• specs.md
• workflows.md
• decisions.md
• gaps.md

Domain Routing

Source topic	Domain
player, renderer, display, window, playback, startup	player-runtime
screen, zone, template, widget, canvas, composer, editor	layout-composition
asset, content, playlist, schedule, campaign, media library	content-assets
source, stream, RTSP, CCTV, M3U, HLS, radio, YouTube, camera	media-sources
menu, POS, order, payment, printing, TV display	restaurant-pos
device, tenant, workspace, onboarding, pairing, fleet, heartbeat	device-fleet
offline, sync, local DB, cache, queue, conflict	sync-offline
event, orchestrator, service registry, workflow, mesh, federation	orchestration
auth, security, role, license, audit, update, secret	security-licensing
AI, generation, guard, optimization, recommendation	ai-optimization
market, strategy, roadmap, Android TV, study, report, pitch	research-strategy

Classification Labels

Use these labels in gaps.md, extraction cards, and backlog updates:

• keep-exact: existing idea is already simple and canonical.
• add-missing: useful idea is absent from canonical docs.
• merge-duplicate: same idea appears in multiple projects.
• expand-needed: idea is promising but incomplete.
• simplify-with-ADR: old logic is complex and a simpler Media OS approach exists.
• obsolete-history: historical detail is not part of current product.
• open-question: product choice cannot be resolved from sources.

Source Reference Standard

Canonical docs cite source folders plus key files. Example:

Source evidence: concepts/screen.md; nexussignage-docs/Composer_Module.md; syngsinc-docs/features/TEMPLATES.md

Do not cite every sentence. Cite major requirements, workflows, decisions, and unresolved gaps.

Processing Workflow

1. Build or update source-manifest.md.
2. Update 01-source-inventory.md with counts, source families, key files, and deferred non-Markdown files.
3. Route Markdown source clusters into domains.
4. For each domain, update README.md, specs.md, workflows.md, decisions.md, and gaps.md.
5. Update overview files after domain processing.
6. Create extraction cards only for high-value, conflicting, duplicate-heavy, or unclear source clusters.

Simplification Rule

When historical projects disagree, prefer the simplest Media OS-compatible approach in canonical specs. Preserve older/complex approaches in ADRs or gaps.md.

Quality Checklist

• Raw source folders are untouched.
• Every domain has the five required files.
• Every domain references at least one source folder and key file.
• Specs use SPEC-* identifiers.
• Mermaid diagrams are fenced with mermaid.
• Non-Markdown files are inventoried before deep extraction.
• Backlog items are actionable and mapped to domains.