From 3-surgeons
HARD-GATE — multi-model review of architectural decisions before implementation proceeds
npx claudepluginhub supportersimulator/3-surgeons --plugin 3-surgeonsThis skill is limited to using the following tools:
This is the primary quality gate for the 3-Surgeons invariance suite. It ensures that no significant change proceeds without proportional scrutiny.
Generates design tokens/docs from CSS/Tailwind/styled-components codebases, audits visual consistency across 10 dimensions, detects AI slop in UI.
Records polished WebM UI demo videos of web apps using Playwright with cursor overlay, natural pacing, and three-phase scripting. Activates for demo, walkthrough, screen recording, or tutorial requests.
Delivers idiomatic Kotlin patterns for null safety, immutability, sealed classes, coroutines, Flows, extensions, DSL builders, and Gradle DSL. Use when writing, reviewing, refactoring, or designing Kotlin code.
This is the primary quality gate for the 3-Surgeons invariance suite. It ensures that no significant change proceeds without proportional scrutiny.
Do NOT implement changes until this gate passes. The gate can be quick (Light mode) or thorough (Full mode), but it MUST run. Skipping this gate violates epistemological invariance.Invoke this skill when ANY of these signals are present:
| Signal | Examples |
|---|---|
| Architectural decision | New subsystem, data flow change, module restructuring |
| >3 files will change | Multi-file refactor, cross-cutting concern |
| New dependency added | npm/pip package, external service, API integration |
| Security-sensitive code | Auth, crypto, input validation, secrets handling |
| Database schema change | Migration, new table, index change, column alter |
| API surface change | New endpoint, modified contract, breaking change |
| Config/infrastructure change | Docker, CI/CD, env vars, deploy config |
If even ONE signal is present, this gate fires.
Not all changes need full gating. The gate assesses risk across multiple dimensions and selects a proportional mode:
| Dimension | Low | Medium | High |
|---|---|---|---|
| Blast radius | 1 file | 2-5 files | >5 files |
| Reversibility | Easy rollback | Moderate effort | Hard/impossible |
| Security exposure | No auth/crypto | Touches validation | Auth/crypto/secrets |
| Data impact | Read-only | Schema additive | Schema destructive |
| External coupling | No new deps | New internal dep | New external service |
| Mode | When | What Runs | Target Time |
|---|---|---|---|
| Light | All dimensions Low. Single-file, clear scope, no arch change | Sentinel scan only. If risk=low, proceed. | <30s |
| Standard | Any dimension Medium. 2-5 files, moderate complexity | Sentinel + gains-gate. Cross-exam if sentinel >= medium. | <120s |
| Full | Any dimension High. >5 files, security, schema, new subsystem | Sentinel + cross-exam + counter-position + gains-gate. | <300s |
Mode is determined by highest-risk dimension, not average. One High dimension = Full mode, even if everything else is Low.
You MUST create a TodoWrite task for each step and complete them in order:
sentinel scan on planned change descriptionsentinel scan on planned change descriptiongains-gate to verify infra health before implementationcross-examination on the plangains-gate again (post-gate)sentinel scan on planned change descriptioncross-examination on the plan (all 3 surgeons)counter-position protocol — argue why this plan will FAILgains-gate to verify infra healthgains-gate again (post-gate)For critical hotfixes (production down, security vulnerability, data loss):
This is the release valve. It exists so the gate does not become the thing that kills the system it protects.
This skill fires BEFORE brainstorming/writing-plans. It supplements the superpowers workflow:
Signal detected → architectural-gate → [brainstorming] → [writing-plans] → [executing-plans]
The gate does NOT replace superpowers' process invariance. It adds epistemological invariance on top — ensuring the right questions are asked before the process begins.