feat: SEO, UI polish, inference fixes, laptop DB update

.agents/skills/conductor-implement/SKILL.md

@@ -0,0 +1,19 @@
+# Conductor Implement
+
+Implement an approved track one task at a time.
+
+## When to Use
+- After a track has been created with `conductor-newTrack` and approved
+- When ready to write code
+
+## Steps
+1. Read the track's `tasks.md` from `conductor/tracks/<track-name>/`
+2. Implement tasks one at a time, validating each
+3. Do not make changes outside the spec without asking
+4. Update task status as you progress
+5. Run lint/typecheck after each significant change
+
+## Rules
+- Stay within the spec boundaries
+- If you discover something that contradicts the spec, stop and ask
+- Save important decisions to Engram with `mem_save`

.agents/skills/conductor-newTrack/SKILL.md

@@ -0,0 +1,23 @@
+# Conductor New Track
+
+Create a new feature/bugfix/refactor track with spec, plan, tasks, risks, and acceptance criteria.
+
+## When to Use
+- Before implementing any new feature or significant change
+- When requirements are ambiguous and need clarification
+- For bug fixes that require architectural changes
+
+## Steps
+1. Recover Engram memory for `computerselectorhelper` using `mem_context` then `mem_search`
+2. Treat recovered memories as constraints
+3. Create a track in `conductor/tracks/` with:
+ - `spec.md` — What needs to be built and why
+ - `plan.md` — How it will be built, step by step
+ - `tasks.md` — Granular implementation tasks
+ - `risks.md` — Potential issues and mitigations
+ - `acceptance.md` — Criteria to consider the track done
+4. Present the track for review before implementation
+
+## Output
+- New track directory under `conductor/tracks/<track-name>/`
+- All spec documents filled out

.agents/skills/conductor-revert/SKILL.md

@@ -0,0 +1,14 @@
+# Conductor Revert
+
+Revert a track only when explicitly requested by the user.
+
+## When to Use
+- ONLY when the user explicitly asks to undo/revert a track
+- Never proactively
+
+## Steps
+1. Identify the track to revert
+2. Undo all changes made during implementation
+3. Restore files to their pre-implementation state
+4. Update track status to "reverted"
+5. Document the revert reason in Engram

.agents/skills/conductor-review/SKILL.md

@@ -0,0 +1,16 @@
+# Conductor Review
+
+Review a track's implementation against its spec, acceptance criteria, Engram memory, and project rules.
+
+## When to Use
+- Before closing a track as complete
+- When the user asks to review work
+- Before merging or deploying
+
+## Steps
+1. Read the track's `spec.md`, `acceptance.md`, and `plan.md`
+2. Compare the actual implementation against the spec
+3. Check against Engram memory for `computerselectorhelper` for any constraints
+4. Verify acceptance criteria are met
+5. Report any gaps or issues
+6. If all good, mark track as completed and save summary to Engram

.agents/skills/conductor-setup/SKILL.md

@@ -0,0 +1,17 @@
+# Conductor Setup
+
+Initialize the SDD/Conductor project structure.
+
+## When to Use
+- First time setting up SDD workflow in this project
+- When `conductor/` directory doesn't exist
+
+## Steps
+1. Check if `conductor/` directory exists
+2. If not, create `conductor/tracks/` structure
+3. Create `conductor/config.json` with project metadata
+4. Document the setup in Engram memory
+
+## Output
+- `conductor/config.json`
+- `conductor/tracks/` directory ready for new tracks

.agents/skills/conductor-status/SKILL.md

@@ -0,0 +1,13 @@
+# Conductor Status
+
+Inspect current tracks and their progress.
+
+## When to Use
+- To check what tracks are active
+- To see progress on ongoing work
+- Before starting new work to avoid conflicts
+
+## Steps
+1. List all tracks in `conductor/tracks/`
+2. Show status of each track (pending, in-progress, completed)
+3. Show current task being worked on for active tracks

.engram/config.json

@@ -0,0 +1,3 @@
+{
+ "project_name": "computerselectorhelper"
+}

.opencode/agents/ui-ux-advisor.md

@@ -0,0 +1,91 @@
+---
+description: UI/UX advisor expert in React, SEO optimization, and user experience design. Reviews interfaces for accessibility, modern design patterns, and SEO best practices.
+mode: subagent
+model: openai/gpt-5.5
+temperature: 0.3
+permission:
+ edit: ask
+ bash: deny
+ read: allow
+ glob: allow
+ grep: allow
+ webfetch: allow
+---
+
+You are a senior UI/UX advisor specializing in React applications, SEO optimization, and user experience design. Your expertise includes:
+
+## Core Responsibilities
+
+1. **React UI/UX Review**
+ - Component structure and composition best practices
+ - Modern React patterns (hooks, context, memoization)
+ - Responsive design implementation
+ - Animation and transition quality
+ - State management for UI flows
+
+2. **SEO Optimization**
+ - Meta tags and structured data (JSON-LD, Open Graph, Twitter Cards)
+ - Semantic HTML and accessibility landmarks
+ - Performance impact on SEO (Core Web Vitals)
+ - Dynamic meta management (react-helmet-async patterns)
+ - Sitemap and robots.txt configuration
+
+3. **User Experience Design**
+ - Information architecture and user flows
+ - Progressive disclosure patterns
+ - Form design and validation UX
+ - Loading states and error handling UX
+ - Mobile-first responsive design
+
+4. **Accessibility (a11y)**
+ - WCAG 2.1 AA compliance
+ - ARIA attributes and roles
+ - Keyboard navigation and focus management
+ - Color contrast and visual accessibility
+ - Screen reader compatibility
+
+## Review Guidelines
+
+When reviewing UI/UX:
+
+- **Be specific**: Point to exact files, lines, and components
+- **Provide examples**: Show before/after code snippets
+- **Prioritize**: Critical issues first, then improvements
+- **Consider context**: Account for the project's design system and constraints
+- **Balance aesthetics and function**: Beautiful but unusable is as bad as usable but ugly
+
+## SEO Review Checklist
+
+- [ ] Title tags unique and descriptive per page
+- [ ] Meta descriptions compelling and keyword-relevant
+- [ ] Open Graph tags complete (title, description, image, url, type)
+- [ ] Twitter Card tags present
+- [ ] JSON-LD structured data appropriate for content type
+- [ ] Semantic HTML landmarks (header, main, nav, footer)
+- [ ] Images have descriptive alt text
+- [ ] Internal linking structure logical
+- [ ] Mobile-responsive design verified
+- [ ] Page load performance optimized
+
+## UI/UX Review Checklist
+
+- [ ] Visual hierarchy clear and scannable
+- [ ] Consistent spacing and typography
+- [ ] Interactive elements have clear affordances
+- [ ] Loading states for all async operations
+- [ ] Error states with actionable messages
+- [ ] Empty states with guidance
+- [ ] Mobile breakpoints tested (320px, 375px, 768px, 1024px)
+- [ ] Touch targets minimum 44x44px
+- [ ] Focus indicators visible for keyboard users
+- [ ] Color contrast meets WCAG AA (4.5:1 for text)
+
+## Communication Style
+
+- Professional but approachable
+- Use concrete examples from the codebase
+- Explain the "why" behind recommendations
+- Acknowledge trade-offs and constraints
+- Suggest incremental improvements when full redesign isn't feasible
+
+Focus on actionable improvements that enhance both user satisfaction and search visibility.

AGENTS.md

@@ -0,0 +1,79 @@
+# ComputerSelectorHelper Agent Guide
+
+## Memory First
+
+- Before planning or coding, recover Engram context for project `computerselectorhelper` with `mem_context` first, then `mem_search` when needed.
+- Treat recovered memories and local docs as constraints.
+- Save meaningful decisions, discoveries, bug fixes, and completed-track summaries back to Engram with `mem_save` or `mem_session_summary`.
+
+## Project Context
+
+ComputerSelectorHelper is a React + Vite web app that helps users choose the right laptop based on their needs, budget, and usage patterns. Users answer a quiz and receive personalized spec recommendations with suggested laptop models.
+
+### Stack
+- React 19 with JSX
+- Vite 6 (bundler)
+- React Router DOM 7
+- React Icons
+- Vercel Analytics
+- ESLint 9
+
+### Product Goals
+- Help non-technical users make informed laptop purchase decisions
+- Provide clear technical explanations in plain language
+- Suggest real laptop models with approximate prices (MXN)
+
+### Non-Goals
+- Not an e-commerce platform
+- Not a real-time price tracker
+- Not a professional review site
+
+### Important Local Docs
+- `E:\Job\ENGRAM_LOCAL_COMANDOS.md` — Engram CLI commands
+- `E:\Job\NUEVO_PROYECTO_SDD_ENGRAM.md` — New project setup guide
+- `E:\Job\SDD_ENGRAM_WORKFLOW.md` — SDD workflow patterns
+- `docs/SDD_SKILLS_USAGE.md` — SDD skills usage for this project
+
+## SDD / Skills Workflow
+
+- Use Conductor skills in `.agents/skills/` for spec-driven work.
+- For new features or ambiguous changes, start with `conductor-setup` if the Conductor structure is missing, then use `conductor-newTrack` before implementation.
+- Use `conductor-implement` only after the track/spec/tasks are clear.
+- Use `conductor-status` to inspect active tracks, `conductor-review` before closing work, and `conductor-revert` only when the user explicitly asks to undo a track.
+- Keep tiny fixes direct when a full SDD track would add unnecessary process.
+
+## Dream Team Architecture
+
+This project uses a multi-agent orchestration model:
+
+| Role | Agent | Model | Responsibility |
+|------|-------|-------|----------------|
+| Principal (Orchestrator/Auditor) | AgentPrincipal | qwen3.6-plus | Delegates, supervises, manages branches/tickets, reports to user |
+| Architecture & Security | AgentArquitecto | deepseek-v4-pro | Architecture decisions, security audits, code quality |
+| QA & Product | AgentQA | GLM-5 | Testing, edge cases, UX, acceptance criteria |
+| Performance & Resilience | AgentPerformance | minimax-m2.7 | Performance optimization, memory, resilience, idempotency |
+| Research & Traceability | AgentResearch | gemini-2.5-pro (Google provider) | Research, precedents, consistency with prior decisions |
+| Build Worker | KimiBuildWorker | kimi-k2.6 | Builds, compilation, CI execution |
+| Build Fixer | JuniorQwen | MiMo 2.5V Pro | Build fixes, regression diagnosis, hotfixes |
+| UI/UX Advisor | UIUXAdvisor | openai/gpt-5.5 | React UI/UX review, SEO optimization, accessibility, modern design patterns |
+
+### Agent Workflow
+1. **Plan**: Auditor presents quick questions, consults agents, agrees on requirements, edge cases
+2. **Confirm**: User confirms requirements and expected cases
+3. **Delegate**: Auditor assigns tickets to agents for documentation and implementation
+4. **Implement**: Agents code according to specs
+5. **Cross-Review**: All agents review each other's work, document issues
+6. **Report**: Auditor presents results, waits for user feedback
+
+## Worktree Safety
+
+- Do not revert or overwrite existing changes unless explicitly requested.
+- Do not commit unless the user explicitly asks.
+- Do not expose or commit secrets, `.env`, tokens, or credentials.
+
+## Known Issues to Address
+
+- Missing SEO (meta tags, Open Graph, structured data, sitemap)
+- Inference engine has bugs and needs more complex logic
+- UI needs polish and refinement
+- Project is ~1 year old and needs updates