domain-overlay

Routes overlapping domain-specific global skills to a single winner per sub-task with stated tie-break reasons.

Updated May 21, 2026
One-click install
npx skills add https://github.com/CagesThrottleUs/private-ai-harness --skill domain-overlay-cagesthrottleus
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: domain-overlay
Source: https://github.com/CagesThrottleUs/private-ai-harness/tree/main/skills/domain-overlay
Command: npx skills add https://github.com/CagesThrottleUs/private-ai-harness --skill domain-overlay-cagesthrottleus

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? When several externally-installed global skills all auto-trigger on the same technology domain (e.g. Android/Kotlin/Compose), they produce contradictory guidance, fire in the wrong order, or crowd out the correct specialist. This Skill replaces fuzzy auto-trigger with deterministic precedence, ordering, and an explicit tie-break reason for every overlap. ## Core Features & Use Cases - Precedence contract: Defines a four-part contract every domain overlay must provide — detection signals, a precedence table with tie-break reasons, lane-step mapping, and a pre-review audit gate. - Reusable precedence heuristics: Supplies decision rules (sole authority, authoring vs measurement, fact-checker role, recency, scope guard) for picking a winner between overlapping skills. - Engineer integration: Hooks into the engineer workflow's construct/test/verify steps so the matching <domain>-advisor decides which specialist fills each existing lane step. - Use Case: A team with four overlapping Compose skills installed creates an android-advisor overlay so that writing a screen deterministically routes to one state-management skill, sequences the perf audit after authoring, and runs a domain self-audit before review. ## Quick Start Ask the agent to apply the domain-overlay pattern to create an advisor for your technology domain that picks a winning skill per sub-task with a stated reason.

Frequently Asked Questions about domain-overlay

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I resolve conflicts between overlapping AI coding skills?

Use a domain overlay that converts fuzzy auto-trigger into deterministic precedence and ordering. For every overlap, pick a winning skill and record a one-line tie-break reason such as sole authority, recency, or scope so the agent stops blending contradictory advice.

What is a domain overlay in an AI agent skill system?

A domain overlay is a thin precedence-and-ordering layer over externally-installed global skills that all trigger on one technology domain. It contains no domain knowledge itself; it routes each sub-task to the correct specialist skill and suppresses conflicting ones.

How do I create a new domain advisor skill for Android or Kotlin?

Copy the four-part contract into a `<domain>-advisor` SKILL.md: detection signals, a precedence table with tie-break reasons, lane-step mapping, and a pre-review audit gate. List installed global skills, pick a winner per overlap, and map survivors to engineer lane steps.

When should I not use a domain overlay?

Skip the overlay when only one skill serves a domain or when installed skills never overlap, since there is nothing to arbitrate. The overlay exists specifically to coordinate multiple domain-deep skills that auto-trigger on the same task.

Does the domain overlay change the engineer workflow steps?

No. The engineer lane structure stays unchanged; the overlay only decides which specialist skill fills each existing construct, test, and verify step. A single Universal-constraints hook activates any matching `<domain>-advisor` by detection signal.