@appsforgood/next-supabase-kit 0.1.4 → 0.1.6

package/examples/next-supabase-installed/audit-output.json

@@ -1,6 +1,6 @@
 {
   "summary": {
-    "pass": 62,
+    "pass": 66,
     "warn": 5,
     "fail": 0
   },
@@ -19,7 +19,7 @@
     {
       "level": "pass",
       "area": "install",
-      "message": "Agent kit installed at version 0.1.4."
+      "message": "Agent kit installed at version 0.1.6."
     },
     {
       "level": "pass",
@@ -96,6 +96,11 @@
       "area": "templates",
       "message": "TESTING.md matches the current bundled template."
     },
+    {
+      "level": "pass",
+      "area": "templates",
+      "message": "LOOP_CODING.md matches the current bundled template."
+    },
     {
       "level": "pass",
       "area": "templates",
@@ -186,6 +191,11 @@
       "area": "agents",
       "message": ".agent-kit/schemas/onboarding-state.schema.json is present and parseable."
     },
+    {
+      "level": "pass",
+      "area": "agents",
+      "message": ".agent-kit/schemas/agentic-level.schema.json is present and parseable."
+    },
     {
       "level": "warn",
       "area": "studio",
@@ -267,6 +277,11 @@
       "area": "docs",
       "message": "TESTING.md exists."
     },
+    {
+      "level": "pass",
+      "area": "docs",
+      "message": "LOOP_CODING.md exists."
+    },
     {
       "level": "pass",
       "area": "docs",
@@ -351,6 +366,11 @@
       "message": "No package.json found to verify test scripts.",
       "remediation": "Add package.json with test, lint, and build scripts appropriate to the stack."
     },
+    {
+      "level": "pass",
+      "area": "project-reality",
+      "message": "No obvious committed secret patterns detected in tracked source files."
+    },
     {
       "level": "pass",
       "area": "project-reality",

package/examples/next-supabase-installed/tree.txt

@@ -14,6 +14,7 @@
 |-- STYLE_GUIDE.md
 |-- SECURITY.md
 |-- TESTING.md
+|-- LOOP_CODING.md
 |-- DEPLOYMENT.md
 |-- UPGRADE.md
 |-- .github

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@appsforgood/next-supabase-kit",
-  "version": "0.1.4",
+  "version": "0.1.6",
   "description": "Open agent council, skills, prompts, checklists, and markdown templates for Next.js and Supabase projects.",
   "type": "module",
   "repository": {
@@ -13,7 +13,8 @@
   "homepage": "https://github.com/lukey662/agentsandskills#readme",
   "publishConfig": {
     "access": "public",
-    "registry": "https://registry.npmjs.org"
+    "registry": "https://registry.npmjs.org",
+    "provenance": true
   },
   "bin": {
     "agent-kit": "dist/index.js"
@@ -57,6 +58,8 @@
     "SUPPLY_CHAIN.md",
     "BEST_PRACTICE_EVIDENCE.md",
     "DOGFOOD.md",
+    "LOOP_CODING.md",
+    "MAINTAINER_RELEASE.md",
     "RESEARCH_CITATION_POLICY.md",
     "UPGRADE.md",
     "SECURITY.md",
@@ -70,8 +73,12 @@
     "build": "tsup && node scripts/copy-studio-assets.mjs",
     "dev": "tsx src/cli/index.ts",
     "typecheck": "tsc --noEmit",
-    "lint": "tsc --noEmit",
+    "lint": "eslint .",
+    "lint:fix": "eslint . --fix",
+    "format": "prettier --write .",
+    "format:check": "prettier --check .",
     "test": "vitest run",
+    "test:coverage": "vitest run --coverage",
     "test:watch": "vitest",
     "smoke": "node dist/index.js doctor",
     "smoke:install": "node scripts/smoke-install.mjs",
@@ -84,20 +91,34 @@
     "sbom:check": "node scripts/sbom-check.mjs",
     "sbom:generate": "node scripts/sbom-check.mjs --output sbom.cdx.json",
     "publish:verify": "node scripts/post-publish-verify.mjs",
+    "changeset": "changeset",
+    "changeset:version": "changeset version",
     "release:check": "node scripts/release-check.mjs",
-    "pack:check": "npm pack --dry-run"
+    "pack:check": "npm pack --dry-run",
+    "dogfood:init": "node scripts/dogfood-init.mjs",
+    "adapter:validate": "node dist/index.js adapter validate all"
   },
   "dependencies": {
-    "@octokit/rest": "^21.1.1",
-    "commander": "^12.1.0",
+    "@clack/prompts": "^1.6.0",
+    "@octokit/rest": "^22.0.1",
+    "commander": "^14.0.3",
+    "picocolors": "^1.1.1",
     "simple-git": "^3.27.0",
-    "zod": "^3.24.1"
+    "zod": "^4.4.3"
   },
   "devDependencies": {
+    "@changesets/cli": "^2.31.0",
+    "@eslint/js": "^9.39.4",
     "@types/node": "^25.9.3",
+    "@vitest/coverage-v8": "^4.1.9",
+    "eslint": "^9.39.4",
+    "eslint-config-prettier": "^10.1.8",
+    "globals": "^17.7.0",
+    "prettier": "^3.9.4",
     "tsup": "^8.3.5",
     "tsx": "^4.19.2",
     "typescript": "^6.0.3",
+    "typescript-eslint": "^8.62.1",
     "vitest": "^4.1.8"
   },
   "overrides": {

package/prompts/ui-command-index.md

@@ -0,0 +1,124 @@
+# UI Command Index
+
+Use this index to run repeatable UI improvement workflows. These are command-like prompts for any assistant surface and the source of truth for native runtime commands.
+
+Canonical sources:
+
+- `DESIGN.md`
+- `STYLE_GUIDE.md`
+- `TESTING.md`
+- `QUALITY_GATES.md`
+- `.agent-kit/agent-roster.json`
+- `.agent-kit/skills/ui-improvement-harness.md`
+- `.agent-kit/checklists/ui-detectors.md`
+- `.agent-kit/checklists/ui-acceptance-rubric.md`
+- `.agent-kit/prompts/screenshot-review.md`
+- `.agent-kit/prompts/visual-qa-plan.md`
+
+## `/ui-audit`
+
+Audit a target UI surface before release.
+
+Required steps:
+
+1. Identify target route, component, workflow, auth state, data state, and risk tier.
+2. Load product/design context and relevant design brief.
+3. Review desktop and mobile evidence, or state which evidence must still be captured.
+4. Run the UI detector checklist.
+5. Return blockers, majors, minors, preserved capabilities, and required fixes.
+
+Required outputs: target surface, risk tier, detector findings by severity, missing evidence, pass/fail verdict.
+
+## `/ui-polish`
+
+Improve visual quality without changing product scope.
+
+Required steps:
+
+1. Start from detector findings, screenshots, or a named surface.
+2. Preserve product behavior, auth boundaries, copy claims, and data contracts.
+3. Improve hierarchy, spacing, density, alignment, component states, and interaction feedback.
+4. Avoid decorative complexity and generic SaaS defaults.
+5. Re-run detector and screenshot critique after changes.
+
+Required outputs: polish goals, changes made or proposed, detector deltas, screenshots needed, residual risks.
+
+## `/layout-cleanup`
+
+Reduce clutter and repair layout structure.
+
+Required steps:
+
+1. Identify primary task, secondary tasks, and content hierarchy.
+2. Remove unnecessary nesting, repeated surfaces, card-within-card patterns, and arbitrary dividers.
+3. Normalize spacing, grouping, alignment, heading scale, and responsive containers.
+4. Verify no truncation, overlap, or horizontal overflow remains.
+
+Required outputs: layout findings, cleanup actions, preserved workflow, desktop/mobile evidence.
+
+## `/responsive-cleanup`
+
+Make a UI work across mobile and desktop.
+
+Required steps:
+
+1. Review mobile width, desktop width, and any tablet breakpoint used by the project.
+2. Verify tap targets, text fit, navigation, primary action visibility, sticky UI, and scroll behavior.
+3. Verify stateful screens on mobile, not only the happy path.
+4. Capture or request mobile and desktop screenshots before acceptance.
+
+Required outputs: viewports checked, responsive findings, fixes, screenshots, remaining risks.
+
+## `/accessibility-pass`
+
+Run a WCAG 2.1 AA-oriented pass on the target UI.
+
+Required steps:
+
+1. Check semantics, headings, landmarks, forms, labels, tables, and buttons.
+2. Check keyboard navigation, visible focus, tab order, escape/close behavior, and skip path when applicable.
+3. Check contrast, reduced motion, error association, touch targets, and status announcements.
+4. Keep auth, state, and data-boundary tests separate from visual checks.
+
+Required outputs: accessibility findings, severity, fixes, skipped checks, test evidence.
+
+## `/distinctiveness-pass`
+
+Prove the UI belongs to this product and does not look generic.
+
+Required steps:
+
+1. Compare the first viewport to product category, audience, workflow, and content fingerprint.
+2. Confirm references were translated into lessons without copying source layouts, copy, assets, or brand marks.
+3. Remove fake metrics, vague claims, abstract filler, and interchangeable card stacks.
+4. Re-score distinctiveness and product quality where significant UI work is involved.
+
+Required outputs: product-specific evidence, generic-risk findings, source-safety notes, required changes.
+
+## `/screenshot-critique`
+
+Review provided screenshots or captured browser screenshots.
+
+Required steps:
+
+1. Name each screenshot, viewport, route, auth state, data state, and UI state.
+2. Run screenshot review, detector checklist, and accessibility risk scan.
+3. Compare against `DESIGN.md`, `STYLE_GUIDE.md`, and selected creative direction.
+4. Return concrete fixes, not broad design advice.
+
+Required outputs: screenshot inventory, blockers, high-value fixes, accepted areas, missing screenshots.
+
+## `/browser-qa`
+
+Run a live browser QA loop for an implemented UI.
+
+Required steps:
+
+1. Start or identify the dev/preview server and target routes.
+2. Open the target with required auth, role, tenant, and data state.
+3. Capture desktop and mobile screenshots.
+4. Run detector, screenshot critique, accessibility pass, and responsive cleanup.
+5. Apply scoped fixes and repeat until acceptance criteria pass.
+6. Record evidence with `agent-kit session output` when Agent Studio is in use.
+
+Required outputs: route, auth state, commands run, screenshots, detector results, fixes, final verdict.

package/research/summaries/agentic-engineering-maturity-levels.md

@@ -0,0 +1,54 @@
+# Agentic Engineering Maturity Levels
+
+Generated from the Agent Kit iceberg model for setup office gamification, `LOOP_CODING.md`, and maintainer climb planning.
+
+## Summary
+
+Agentic engineering maturity is modeled as **levels L3–L8**. The kit **computes L3–L6** from audit, adapter, and setup signals. **L7–L8 are deferred** (never auto-awarded) because they require stronger eval gates than agent freedom.
+
+| Level | Name | Computed? |
+| --- | --- | --- |
+| L3 | AI-native IDE | Yes |
+| L4 | Rules, context, MCP contract | Yes |
+| L5 | Subagents, skills, council specialists | Yes |
+| L6 | Eval loops, hooks, CI gates | Yes (downstream or maintainer profile) |
+| L7 | Overnight / unsupervised teams | No — explore only |
+| L8 | Agents managing agents | No — lab only |
+
+## Base levels (reference)
+
+| Lens | Level |
+| --- | --- |
+| What `@appsforgood/next-supabase-kit` ships | High L4 → solid L5 |
+| BaseRepo maintainer day-to-day (gitignored dogfood overlay) | Mid L4 |
+| BaseRepo CI (`release:check`, `adapter:validate`, audit smokes) | Partial L6 |
+| Fresh downstream `init` | ~L4 capability; L3 until context complete |
+| Fresh `init --activate cursor\|codex` | L5 capability when subagents exist |
+
+## When each level was cutting edge (approximate)
+
+| Era | Frontier | Became baseline |
+| --- | --- | --- |
+| 2022–2023 | Browser chat, tab completion | Copy-paste only |
+| 2023–2024 | Repo-aware IDE chat (L3) | Autocomplete-only |
+| Late 2024–2025 | AGENTS.md, rules, MCP (L4) | Ad-hoc prompting |
+| 2025–2026 | Subagents, skills, councils (L5) | Single chat for all roles |
+| 2026 now | Eval CI, loop until green (L6) | Manual “run tests when you remember” |
+| 2026 hype | Overnight agent teams (L7) | Not safe as default |
+| 2026+ lab | Meta-orchestration (L8) | Not mainstream production |
+
+## Office and wizard integration
+
+- `agent-kit setup` computes level via `computeAgenticLevel()` and returns it on `/api/state`.
+- Agent Office shows **Lcurrent → Ltarget**, an iceberg strip (L3–L8), and top climb steps.
+- Setup wizard complete screen links **`LOOP_CODING.md`**, **`agent-kit adapter validate`**, and audit gates.
+- Maintainer source repos use a **maintainer profile** (release-check + dogfood docs) for L6 signals.
+
+## Related assets
+
+- `LOOP_CODING.md` — kit-safe loop patterns
+- `schemas/agentic-level.schema.json` — report shape
+- `src/studio/agentic-level.ts` — scoring implementation
+- `research/summaries/maturity-model-patterns.md` — QUALITY_GATES evidence tiers (separate from Agentic L3–L8)
+
+Do not conflate **Agentic L3–L8** with **QUALITY_GATES Baseline/Strong/Best-Practice** or **visual QA baseline/strong/mature** — the setup UI labels these separately.

package/rosters/next-supabase-default-council.json

@@ -70,6 +70,7 @@
         "frontend-distinctiveness-benchmark",
         "frontend-product-quality-rubric",
         "frontend-design-system",
+        "ui-improvement-harness",
         "visual-regression-qa",
         "accessibility-wcag"
       ],
@@ -95,13 +96,7 @@
         "voice-tone",
         "pricing-copy"
       ],
-      "skills": [
-        "positioning-messaging",
-        "conversion-copywriting",
-        "landing-page-copy",
-        "product-voice-tone",
-        "onboarding-empty-state-copy"
-      ],
+      "skills": ["positioning-messaging", "conversion-copywriting", "landing-page-copy", "product-voice-tone", "onboarding-empty-state-copy"],
       "handsOffTo": ["frontend-design-lead", "nextjs-engineer", "qa-engineer", "docs-maintainer"]
     },
     {
@@ -147,7 +142,19 @@
     },
     {
       "id": "core-change",
-      "triggers": ["schema", "auth", "rls", "api", "route handler", "server action", "dependency", "upgrade", "release workflow", "package behavior", "cross-layer"],
+      "triggers": [
+        "schema",
+        "auth",
+        "rls",
+        "api",
+        "route handler",
+        "server action",
+        "dependency",
+        "upgrade",
+        "release workflow",
+        "package behavior",
+        "cross-layer"
+      ],
       "sequence": [
         "planner",
         "lead-architect",
@@ -171,7 +178,7 @@
     },
     {
       "id": "frontend-change",
-      "triggers": ["screen", "component", "layout", "design", "responsive", "accessibility", "screenshot"],
+      "triggers": ["screen", "component", "layout", "design", "responsive", "accessibility", "screenshot", "ui audit", "ui polish", "browser qa"],
       "sequence": ["planner", "frontend-design-lead", "marketing-copy-lead", "nextjs-engineer", "qa-engineer", "docs-maintainer"],
       "council": ["frontend-design-lead", "marketing-copy-lead", "qa-engineer"],
       "requiredOutputs": [
@@ -183,15 +190,33 @@
         "design critique verdict",
         "frontend product-quality scorecard",
         "domain-specific UI rationale",
+        "UI detector findings and severity",
+        "UI command workflow applied when polishing or auditing",
         "visual QA evidence",
         "state coverage",
         "accessibility checks",
-        "desktop/mobile verification"
+        "desktop/mobile verification",
+        "authenticated screen evidence when applicable"
       ]
     },
     {
       "id": "marketing-copy",
-      "triggers": ["copy", "copywriting", "marketing", "positioning", "messaging", "value prop", "value proposition", "landing page", "headline", "cta", "conversion", "onboarding", "empty state", "pricing"],
+      "triggers": [
+        "copy",
+        "copywriting",
+        "marketing",
+        "positioning",
+        "messaging",
+        "value prop",
+        "value proposition",
+        "landing page",
+        "headline",
+        "cta",
+        "conversion",
+        "onboarding",
+        "empty state",
+        "pricing"
+      ],
       "sequence": ["planner", "marketing-copy-lead", "frontend-design-lead", "qa-engineer", "docs-maintainer"],
       "council": ["marketing-copy-lead", "frontend-design-lead"],
       "requiredOutputs": [
@@ -219,7 +244,7 @@
     "Meaningful multi-agent work must record council-session evidence in COUNCIL.md or a structured record that follows .agent-kit/schemas/council-session.schema.json.",
     "Planner starts planning and ambiguous requests by default.",
     "Lead Architect must review core changes before implementation.",
-    "Frontend Design Lead must record reference-set evidence, anti-references, a design critique verdict, a distinctiveness benchmark, and a frontend product-quality scorecard before accepting significant frontend work.",
+    "Frontend Design Lead must record reference-set evidence, anti-references, a design critique verdict, a distinctiveness benchmark, a frontend product-quality scorecard, UI detector severity findings, and desktop/mobile screenshot evidence before accepting significant frontend work.",
     "Marketing Copy Lead must ask discovery questions and record audience, pain, outcome, value proposition, proof, objections, voice/tone, and conversion goal before accepting public-facing or conversion-facing copy.",
     "Security Reviewer must review auth, data mutation, external-call, dependency, secret, and release-risk changes.",
     "QA Engineer must verify behavior changes before completion.",

package/runtime-skills/ui-improvement-harness/SKILL.md

@@ -0,0 +1,12 @@
+---
+name: ui-improvement-harness
+description: Use when running repeatable UI audit, polish, layout cleanup, responsive cleanup, accessibility, screenshot critique, visual distinctiveness, or live browser QA workflows.
+---
+
+# UI Improvement Harness Runtime Skill
+
+Canonical workflow: `skills/ui-improvement-harness.md`.
+
+Installed-project canonical workflow: `.agent-kit/skills/ui-improvement-harness.md`.
+
+Use the canonical skill before accepting meaningful UI polish, audit, screenshot review, or browser iteration work. Runtime wrappers must not fork the detector, rubric, or command policy.

package/schemas/agentic-level.schema.json

@@ -0,0 +1,47 @@
+{
+  "$schema": "https://json-schema.org/draft/2020-12/schema",
+  "$id": "https://agent-skills.dev/schemas/agentic-level.schema.json",
+  "title": "Agent Kit Agentic Level Report",
+  "type": "object",
+  "required": ["currentLevel", "targetLevel", "maintainerProfile", "computedAt", "signals"],
+  "additionalProperties": false,
+  "properties": {
+    "currentLevel": { "type": "integer", "enum": [3, 4, 5, 6] },
+    "targetLevel": { "type": "integer", "enum": [3, 4, 5, 6, 7, 8] },
+    "maintainerProfile": { "type": "boolean" },
+    "computedAt": { "type": "string", "format": "date-time" },
+    "maintainerNote": { "type": "string" },
+    "signals": {
+      "type": "array",
+      "items": {
+        "type": "object",
+        "required": ["id", "level", "label", "pass", "evidence", "remediation"],
+        "additionalProperties": false,
+        "properties": {
+          "id": { "type": "string" },
+          "level": { "type": "integer", "enum": [3, 4, 5, 6] },
+          "label": { "type": "string" },
+          "pass": { "type": "boolean" },
+          "evidence": { "type": "string" },
+          "remediation": { "type": "string" }
+        }
+      }
+    },
+    "climbSteps": {
+      "type": "array",
+      "items": {
+        "type": "object",
+        "required": ["id", "level", "label", "pass", "evidence", "remediation"],
+        "additionalProperties": false,
+        "properties": {
+          "id": { "type": "string" },
+          "level": { "type": "integer", "enum": [3, 4, 5, 6] },
+          "label": { "type": "string" },
+          "pass": { "type": "boolean" },
+          "evidence": { "type": "string" },
+          "remediation": { "type": "string" }
+        }
+      }
+    }
+  }
+}

package/schemas/onboarding-state.schema.json

@@ -28,6 +28,9 @@
     "wizardVersion": { "type": "string" },
     "ideSurface": { "type": "string", "enum": ["cursor", "copilot", "claude", "codex", "other"] },
     "ideVerifiedAt": { "type": "string", "format": "date-time" },
-    "visualQaTier": { "type": "string", "enum": ["baseline", "strong", "mature"] }
+    "visualQaTier": { "type": "string", "enum": ["baseline", "strong", "mature"] },
+    "targetAgenticLevel": { "type": "integer", "enum": [3, 4, 5, 6, 7, 8] },
+    "lastAgenticLevel": { "type": "integer", "enum": [3, 4, 5, 6] },
+    "lastAgenticComputedAt": { "type": "string", "format": "date-time" }
   }
 }

package/skills/ui-improvement-harness.md

@@ -0,0 +1,96 @@
+# UI Improvement Harness Skill
+
+## Use When
+
+Running operational UI improvement loops for user-facing screens, dashboards, admin flows, app shells, onboarding, marketing surfaces, component systems, or authenticated product workflows.
+
+Use this skill when a request asks for UI audit, polish, layout cleanup, responsive cleanup, accessibility pass, screenshot critique, visual distinctiveness, live browser QA, or repeatable frontend improvement workflows.
+
+## Goal
+
+Turn frontend quality guidance into repeatable commands. The harness should help agents move from screenshots and browser evidence to concrete fixes, then loop until deterministic detector findings and review findings pass for the release risk.
+
+## Required Inputs
+
+- Target surface: route, component, flow, or screenshot set.
+- Product context from `DESIGN.md`, `STYLE_GUIDE.md`, and relevant `.agent-kit/design-briefs/*`.
+- Auth and permission context for protected screens, including role, tenant, and fixture user assumptions.
+- Desktop and mobile viewports for meaningful UI changes.
+- State coverage requirements: default, loading, empty, error, disabled, success, permission, focus, and narrow viewport states where relevant.
+- Existing visual QA tier from `TESTING.md`, or a proposed tier from `.agent-kit/prompts/visual-qa-plan.md`.
+
+## Command Workflows
+
+Use `.agent-kit/prompts/ui-command-index.md` as the command index. It defines these command-like workflows:
+
+- UI audit: find blockers and major issues before implementation or release.
+- UI polish: improve hierarchy, density, spacing, state feedback, and visual finish without changing product scope.
+- Layout cleanup: remove clutter, nesting, overflow, weak grouping, and inconsistent spacing.
+- Responsive/mobile cleanup: verify mobile hierarchy, tap targets, text fit, navigation, and critical actions.
+- Accessibility pass: verify semantic structure, keyboard path, focus, labels, contrast, motion, and error feedback.
+- Visual distinctiveness pass: prove product specificity and avoid generic AI/SaaS defaults.
+- Screenshot critique: review desktop/mobile/state screenshots against product context and detector findings.
+- Live browser QA loop: inspect the running app, capture evidence, apply scoped fixes, and repeat until acceptance criteria pass.
+
+## Deterministic Detector Checklist
+
+Use `.agent-kit/checklists/ui-detectors.md` before and after polish. Treat each finding as `blocker`, `major`, `minor`, or `not-applicable`.
+
+Required detector areas:
+
+- Crowded layout and excessive competing surfaces.
+- Weak hierarchy or unclear primary action.
+- Inconsistent spacing, density, typography, radius, or alignment.
+- Poor mobile tap targets or unreachable critical actions.
+- Truncation, overflow, overlap, or text that does not fit.
+- Card-within-card patterns and decorative containers that obscure workflow.
+- Generic SaaS or AI-site styling, including abstract decoration that hides missing content.
+- Weak hover, pressed, selected, loading, success, empty, error, disabled, focus, and permission feedback.
+- Missing loading, empty, error, disabled, success, permission, and mobile states where relevant.
+- Accessibility and focus issues, including contrast, labels, landmarks, keyboard order, and motion.
+
+## Acceptance Rubric
+
+Use `.agent-kit/checklists/ui-acceptance-rubric.md`.
+
+- Blocker findings must be fixed before release.
+- Major findings must be fixed or explicitly accepted with rationale before high-risk UI changes ship.
+- High-risk UI changes require desktop and mobile screenshots plus authenticated or permission-state evidence when the workflow requires login.
+- Screenshot evidence must include the first viewport, primary workflow, and the highest-risk state affected by the change.
+- Visual checks do not replace semantic, keyboard, auth, data-boundary, or regression tests.
+
+## Live Browser QA Loop
+
+For implemented UI:
+
+1. Start or identify the app server and route.
+2. Open the target route with the required auth state or test account.
+3. Capture desktop and mobile screenshots.
+4. Run the deterministic detector checklist.
+5. Run screenshot critique and accessibility pass.
+6. Apply the smallest scoped fixes.
+7. Repeat until no blockers remain and major findings are resolved or documented.
+8. Record evidence with `agent-kit session output` when Agent Studio is in use.
+
+For authenticated screens, do not substitute a public landing page screenshot. Use a real signed-in state, a seeded local fixture, or an explicit limitation note that names the missing auth evidence.
+
+## Reject By Default
+
+- A UI pass based only on static prose with no screenshot or browser evidence when implementation exists.
+- Desktop-only review for responsive or mobile-risk changes.
+- Public-page review used as evidence for authenticated app screens.
+- Vague polish requests that do not name detector findings or acceptance criteria.
+- Fixes that add decorative cards, gradients, fake metrics, or generic copy instead of clarifying the workflow.
+- Ignoring major detector findings without rationale.
+
+## Review Output
+
+Return:
+
+- Target surface and auth state reviewed.
+- Commands run or command-like workflows applied.
+- Detector findings by severity.
+- Screenshots and viewports reviewed.
+- Accessibility, responsive, state, and distinctiveness verdicts.
+- Fixes made or required before acceptance.
+- Tests, browser checks, visual QA evidence, and gaps.

package/templates/next-supabase/AGENT_ROSTER.md

@@ -4,7 +4,7 @@ This project uses `.agent-kit/agent-roster.json` as the default council contract
 
 ## Default Rule
 
-Planner handles planning by default. Lead Architect reviews core changes before implementation. Frontend Design Lead owns content-first creative direction, reference-led critique, frontend distinctiveness benchmarking, product-quality scoring, and visual QA before significant frontend implementation is accepted. Marketing Copy Lead owns positioning, value proposition, public-facing copy, proof, objections, voice, and CTA hierarchy before conversion-facing copy is accepted. Security Reviewer, QA Engineer, Documentation Maintainer, and Deployment/Observability Engineer join when their trigger areas are touched. Meaningful multi-agent work records council-session evidence in `COUNCIL.md` or a structured record that follows `.agent-kit/schemas/council-session.schema.json`.
+Planner handles planning by default. Lead Architect reviews core changes before implementation. Frontend Design Lead owns content-first creative direction, reference-led critique, frontend distinctiveness benchmarking, product-quality scoring, UI detector severity review, command-based UI polish/audit loops, and visual QA before significant frontend implementation is accepted. Marketing Copy Lead owns positioning, value proposition, public-facing copy, proof, objections, voice, and CTA hierarchy before conversion-facing copy is accepted. Security Reviewer, QA Engineer, Documentation Maintainer, and Deployment/Observability Engineer join when their trigger areas are touched. Meaningful multi-agent work records council-session evidence in `COUNCIL.md` or a structured record that follows `.agent-kit/schemas/council-session.schema.json`.
 
 ## Default Workflows
 
@@ -36,7 +36,7 @@ Handoff order:
 
 ### Frontend Change
 
-Use when the request touches screens, components, layout, visual design, accessibility, responsiveness, or screenshot review.
+Use when the request touches screens, components, layout, visual design, accessibility, responsiveness, screenshot review, UI audit, UI polish, or browser QA.
 
 Handoff order:
 
@@ -57,10 +57,13 @@ Required outputs:
 - Design critique verdict
 - Frontend product-quality scorecard
 - Domain-specific UI rationale
+- UI detector findings and severity
+- UI command workflow applied when polishing or auditing
 - Visual QA evidence
 - State coverage
 - Accessibility checks
 - Desktop/mobile verification
+- Authenticated screen evidence when applicable
 
 ### Marketing Copy
 
@@ -90,7 +93,7 @@ Required outputs:
 - Each agent must state its decision, risk, and required next handoff.
 - Each meaningful council session must record workflow, affected layers, required outputs, handoff decisions, risks, evidence, and verification status.
 - Core changes cannot skip Lead Architect.
-- Frontend changes cannot skip content/brand intake, creative-direction rationale, reference-set evidence, distinctiveness benchmark, design critique verdict, product-quality scorecard, visual QA evidence, or Frontend Design Lead review.
+- Frontend changes cannot skip content/brand intake, creative-direction rationale, reference-set evidence, distinctiveness benchmark, design critique verdict, product-quality scorecard, UI detector severity review, visual QA evidence, or Frontend Design Lead review.
 - Public-facing or conversion-facing copy cannot skip Marketing Copy Lead discovery questions, value proposition, proof, objection, voice/tone, and CTA review.
 - Auth, data mutation, dependency, external-call, secret, and release-risk changes cannot skip Security Reviewer.
 - Behavior changes cannot skip QA evidence.

package/templates/next-supabase/ASSISTANT_ADAPTERS.md

@@ -108,7 +108,9 @@ Validate the structural package with:
 agent-kit adapter validate antigravity
 ```
 
-If the local machine has the Antigravity CLI, teams may also run native plugin validation and record that evidence here. Runtime command files are adapters only; `AGENTS.md`, `.agent-kit/agent-roster.json`, `QUALITY_GATES.md`, and Agent Studio sessions remain canonical.
+If the local machine has the Antigravity CLI, teams may also run native plugin validation and record that evidence here. Runtime command files are adapters only; `AGENTS.md`, `.agent-kit/agent-roster.json`, `QUALITY_GATES.md`, `.agent-kit/prompts/ui-command-index.md`, `.agent-kit/checklists/ui-detectors.md`, and Agent Studio sessions remain canonical.
+
+The frontend adapter surface includes `/frontend` plus focused UI improvement commands: `/ui-audit`, `/ui-polish`, `/layout-cleanup`, `/responsive-cleanup`, `/accessibility-pass`, `/distinctiveness-pass`, `/screenshot-critique`, and `/browser-qa`. These commands invoke existing Frontend Design Lead, QA, accessibility, screenshot, and documentation gates; they do not replace the council workflow.
 
 ## Acceptance Evidence