skill-domain-routing-designer
Skill Domain Routing (theonekit-designer)
Section titled “Skill Domain Routing (theonekit-designer)”The mechanics/economy/feel/narrative/UI-UX/documentation intent→skill lookup tables (moved
2026-08-24, context-budget lever — pure reference consulted only after the agent already decided
to do design work) now live in the on-demand skill t1k-designer-base-skill-domain-routing. Invoke
it — or /t1k:find-skill "<query>" — when resolving “which t1k-designer-* skill handles X.”
For core T1K skills (cook, fix, plan, etc.), see skills/t1k-help/references/skill-domain-routing.md.
Review & Validation (Design-Perspective, NOT Code-Reviewer)
Section titled “Review & Validation (Design-Perspective, NOT Code-Reviewer)”This stays a rule (not the skill above) because it is a prohibition that fires unsummoned — it prevents an agent from reaching for t1k-code-reviewer before it would ever think to look up a routing table.
Rule: Reviewing or validating design docs, wiki pages, narrative content, balance plans, or any artifact produced by a designer-kit skill MUST use design-perspective agents — NOT t1k-code-reviewer. That agent is for code-quality concerns (security, patterns, DRY/KISS); it is the wrong lens for design artifacts.
User wants to…
- Validate plan completeness, phase structure, dependency graph, success criteria →
t1k-planneragent (core; audits its own discipline) - Review design docs for content/pillar coverage — does the deliverable serve the documented design pillars? →
t1k-designer-game-designeragent - Review design docs for data SSOT compliance — is narrative referencing CSV/JSON source-of-truth correctly? Activate
t1k:designer:base:data-authoringskill alongside →t1k-designer-game-designeragent - Review for tone, voice, pillar fidelity, alternative-tone risks during multi-agent fan-out →
t1k-designer-brainstormeragent - Review for production readiness, ship-gate, stakeholder-decision propagation, milestone gates →
t1k-designer-game-produceragent - Review for wiki organization, navigability, cross-link consistency →
t1k-designer-game-designeragent (witht1k:designer:wiki:coreskill active)
Multi-round design validation (5-round adversarial pattern)
Section titled “Multi-round design validation (5-round adversarial pattern)”When the user requests “validate in N rounds” or asks for adversarial review on design artifacts, prefer this distribution over generic t1k-code-reviewer chains:
| Round | Agent | Lens |
|---|---|---|
| 1 | t1k-planner | Plan rigor — phase structure, file ownership, success criteria coverage, risk register |
| 2 | t1k-designer-game-designer | Design completeness — pillars served, content coverage, no contradictions to GDD |
| 3 | t1k-designer-game-designer (with t1k:designer:base:data-authoring) | Data SSOT — CSV/JSON read-before-write, no narrative-data drift |
| 4 | t1k-designer-brainstormer | Tone & voice — pillar fidelity under parallel fan-out, alternative-tone risks |
| 5 | t1k-designer-game-producer | Production readiness — stakeholder gates, ship gates, milestone alignment |
Each name above is the exact subagent_type to pass to Agent — the identity, per rules/agent-name-is-identity.md. t1k-planner ships with core; the other three ship with this kit.
Anti-pattern: spawning ≥2 t1k-code-reviewer agents for design-doc validation. It cannot judge whether lore content serves the game’s pillars, whether tooltips drift from items.csv, or whether a story bible’s faction conflict reads as comedy vs. melodrama. Use designer-perspective agents.