wegho-agentes 7.0.4 → 7.0.6

package/skills/loki-mode/demo/voice-over-script.md

@@ -1,246 +0,0 @@
-# Loki Mode Voice-Over Script
-
-Complete narration for Loki Mode demo video.
-
----
-
-## Introduction (0:00 - 0:30)
-
-> Welcome to Loki Mode - a multi-agent autonomous startup system for Claude Code.
->
-> Loki Mode takes your product requirements document and transforms it into a fully functioning application - with zero human intervention.
->
-> Today I'll show you how it works by building a complete todo application from scratch.
-
----
-
-## Setup (0:30 - 1:00)
-
-> First, we launch Claude Code with the dangerously-skip-permissions flag. This allows Loki Mode to run autonomously without asking for confirmation at every step.
->
-> [Show terminal: `claude --dangerously-skip-permissions`]
->
-> Now we invoke Loki Mode with our PRD.
-
----
-
-## Invocation (1:00 - 1:30)
-
-> [Type: "Loki Mode with PRD at examples/simple-todo-app.md"]
->
-> Loki Mode immediately begins the RARV cycle - Reason, Act, Reflect, Verify.
->
-> It first reads the PRD to understand what we're building.
-
----
-
-## Bootstrap Phase (1:30 - 2:30)
-
-> Notice Loki Mode is now in the Bootstrap phase. It's setting up the project structure.
->
-> [Show: .loki directory being created]
->
-> The .loki directory contains:
-> - CONTINUITY.md - the working memory that persists across context resets
-> - Queue files for task management
-> - State tracking for the orchestrator
->
-> This is how Loki Mode maintains context even during long-running operations.
-
----
-
-## Discovery Phase (2:30 - 3:30)
-
-> Now we're in Discovery. Loki Mode is analyzing our PRD and extracting requirements.
->
-> [Show: Tasks being generated]
->
-> See how it breaks down the todo app into specific tasks:
-> - Set up backend with Express
-> - Create SQLite database schema
-> - Implement API endpoints
-> - Build React frontend
->
-> Each task gets added to the pending queue.
-
----
-
-## Architecture Phase (3:30 - 4:30)
-
-> The Architecture phase is where Loki Mode designs the system.
->
-> [Show: OpenAPI spec being created]
->
-> Notice it's following spec-first development - the OpenAPI specification is created BEFORE any code is written.
->
-> This ensures the frontend and backend will work together seamlessly.
-
----
-
-## Kanban Visualization (4:30 - 5:30)
-
-> Let me show you the Vibe Kanban integration.
->
-> [Show: Kanban board with tasks]
->
-> Each task appears on our kanban board. As agents claim tasks, they move from "To Do" to "In Progress" to "Done".
->
-> This gives you real-time visibility into what Loki Mode is doing.
-
----
-
-## Agent Spawning (5:30 - 7:00)
-
-> Now watch the magic happen.
->
-> [Show: Multiple agents being spawned]
->
-> Loki Mode spawns specialized agents:
-> - A backend agent implementing the Express server
-> - A frontend agent building the React UI
-> - A database agent setting up SQLite
->
-> These agents work in parallel - but notice they're not stepping on each other's toes. The task queue system prevents conflicts.
-
----
-
-## Model Selection (7:00 - 7:30)
-
-> Pay attention to the model selection.
->
-> Simple tasks like running tests use Haiku - fast and cost-effective.
-> Standard implementation uses Sonnet - the default workhorse.
-> Complex decisions like architecture use Opus - for deep analysis.
->
-> This intelligent routing optimizes both speed and quality.
-
----
-
-## Code Review (7:30 - 9:00)
-
-> Here's my favorite part - the code review system.
->
-> [Show: Three reviewers being dispatched]
->
-> Loki Mode dispatches THREE reviewers in parallel:
-> 1. Code quality reviewer - checks patterns and best practices
-> 2. Business logic reviewer - verifies requirements are met
-> 3. Security reviewer - scans for vulnerabilities
->
-> They review independently - blind to each other's findings. This prevents groupthink.
->
-> [Show: Review results]
->
-> If all three approve, a Devil's Advocate reviewer is triggered. This fourth reviewer specifically looks for issues the others might have missed.
->
-> This anti-sycophancy protocol catches 30% more issues than traditional reviews.
-
----
-
-## Quality Gates (9:00 - 10:00)
-
-> Severity-based blocking ensures nothing ships broken.
->
-> [Show: Quality gate output]
->
-> Critical, High, and Medium issues BLOCK the pipeline.
-> Low and Cosmetic issues get TODO comments but don't block.
->
-> Tests must pass. Coverage must exceed 80%. No exceptions.
-
----
-
-## CONTINUITY.md (10:00 - 11:00)
-
-> Let's peek at the working memory.
->
-> [Show: CONTINUITY.md contents]
->
-> This file tracks:
-> - Current task and progress
-> - Decisions made and why
-> - Mistakes and learnings
->
-> If Loki Mode runs out of context or needs to restart, it reads this file first. This is how it maintains coherence across long sessions.
-
----
-
-## Memory System (11:00 - 12:00)
-
-> Loki Mode has a three-layer memory system.
->
-> Episodic memory records what happened - specific actions and their outcomes.
->
-> Semantic memory generalizes patterns - "TypeScript strict mode requires explicit return types."
->
-> Procedural memory stores learned skills - how to implement an API endpoint successfully.
->
-> This isn't just context - it's genuine learning that improves future runs.
-
----
-
-## Completion (12:00 - 13:00)
-
-> [Show: Application running]
->
-> And here's our finished todo app!
->
-> - Full CRUD operations working
-> - React frontend with TypeScript
-> - Express backend with SQLite
-> - All tests passing
-> - Code reviewed and approved
->
-> From PRD to working application - completely autonomous.
-
----
-
-## Recap (13:00 - 14:00)
-
-> Let's recap what Loki Mode did:
->
-> 1. Read and analyzed the PRD
-> 2. Designed the architecture with OpenAPI specs
-> 3. Spawned specialized agents for parallel development
-> 4. Ran comprehensive code reviews with anti-sycophancy checks
-> 5. Enforced quality gates and test coverage
-> 6. Maintained context through the memory system
->
-> All without a single human intervention.
-
----
-
-## Call to Action (14:00 - 14:30)
-
-> Loki Mode is available now on GitHub.
->
-> Install it as a Claude Code skill and start building.
->
-> Remember to use the dangerously-skip-permissions flag for full autonomy.
->
-> Thanks for watching!
-
----
-
-## Timing Summary
-
-| Section | Start | Duration |
-|---------|-------|----------|
-| Introduction | 0:00 | 30s |
-| Setup | 0:30 | 30s |
-| Invocation | 1:00 | 30s |
-| Bootstrap | 1:30 | 60s |
-| Discovery | 2:30 | 60s |
-| Architecture | 3:30 | 60s |
-| Kanban | 4:30 | 60s |
-| Agents | 5:30 | 90s |
-| Model Selection | 7:00 | 30s |
-| Code Review | 7:30 | 90s |
-| Quality Gates | 9:00 | 60s |
-| CONTINUITY | 10:00 | 60s |
-| Memory | 11:00 | 60s |
-| Completion | 12:00 | 60s |
-| Recap | 13:00 | 60s |
-| CTA | 14:00 | 30s |
-
-**Total: ~14.5 minutes**

package/skills/loki-mode/examples/api-only.md

@@ -1,79 +0,0 @@
-# PRD: REST API Service
-
-## Overview
-A simple REST API for managing notes. Tests Loki Mode's backend-only capabilities.
-
-## Target Users
-Developers who need a notes API.
-
-## API Endpoints
-
-### Notes Resource
-
-#### GET /api/notes
-- Returns list of all notes
-- Response: `[{ id, title, content, createdAt }]`
-
-#### GET /api/notes/:id
-- Returns single note
-- Response: `{ id, title, content, createdAt }`
-- Error: 404 if not found
-
-#### POST /api/notes
-- Creates new note
-- Body: `{ title, content }`
-- Response: `{ id, title, content, createdAt }`
-- Error: 400 if validation fails
-
-#### PUT /api/notes/:id
-- Updates existing note
-- Body: `{ title?, content? }`
-- Response: `{ id, title, content, updatedAt }`
-- Error: 404 if not found
-
-#### DELETE /api/notes/:id
-- Deletes note
-- Response: 204 No Content
-- Error: 404 if not found
-
-### Health Check
-
-#### GET /health
-- Returns `{ status: "ok", timestamp }`
-
-## Tech Stack
-- Runtime: Node.js 18+
-- Framework: Express.js
-- Database: In-memory (array) for simplicity
-- Validation: zod or joi
-- Testing: Jest + supertest
-
-## Requirements
-- Input validation on all endpoints
-- Proper HTTP status codes
-- JSON error responses
-- Request logging
-- Unit tests for each endpoint
-
-## Out of Scope
-- Authentication
-- Database persistence
-- Rate limiting
-- API documentation (OpenAPI)
-- Deployment
-
-## Test Cases
-```
-POST /api/notes with valid data → 201 + note object
-POST /api/notes with missing title → 400 + error
-GET /api/notes → 200 + array
-GET /api/notes/:id with valid id → 200 + note
-GET /api/notes/:id with invalid id → 404
-PUT /api/notes/:id with valid data → 200 + updated note
-DELETE /api/notes/:id → 204
-GET /health → 200 + status object
-```
-
----
-
-**Purpose:** Tests backend agent capabilities, code review, and QA without frontend complexity.

package/skills/loki-mode/examples/full-stack-demo.md

@@ -1,123 +0,0 @@
-# PRD: Full-Stack Demo App
-
-## Overview
-A complete full-stack application demonstrating Loki Mode's end-to-end capabilities. A simple bookmark manager with tags.
-
-## Target Users
-Users who want to save and organize bookmarks.
-
-## Features
-
-### Core Features
-1. **Add Bookmark** - Save URL with title and optional tags
-2. **View Bookmarks** - List all bookmarks with search/filter
-3. **Edit Bookmark** - Update title, URL, or tags
-4. **Delete Bookmark** - Remove bookmark
-5. **Tag Management** - Create, view, and filter by tags
-
-### User Flow
-1. User opens app → sees bookmark list
-2. Clicks "Add Bookmark" → form appears
-3. Enters URL, title, tags → submits
-4. Bookmark appears in list
-5. Can filter by tag or search by title
-6. Can edit or delete any bookmark
-
-## Tech Stack
-
-### Frontend
-- React 18 with TypeScript
-- Vite for bundling
-- TailwindCSS for styling
-- React Query for data fetching
-
-### Backend
-- Node.js 18+
-- Express.js
-- SQLite with better-sqlite3
-- zod for validation
-
-### Structure
-```
-/
-├── frontend/
-│   ├── src/
-│   │   ├── components/
-│   │   ├── hooks/
-│   │   ├── types/
-│   │   └── App.tsx
-│   ├── package.json
-│   └── vite.config.ts
-├── backend/
-│   ├── src/
-│   │   ├── routes/
-│   │   ├── db/
-│   │   └── index.ts
-│   ├── package.json
-│   └── tsconfig.json
-└── README.md
-```
-
-## API Endpoints
-
-### Bookmarks
-- `GET /api/bookmarks` - List all (query: `?tag=`, `?search=`)
-- `POST /api/bookmarks` - Create new
-- `PUT /api/bookmarks/:id` - Update
-- `DELETE /api/bookmarks/:id` - Delete
-
-### Tags
-- `GET /api/tags` - List all tags with counts
-
-## Database Schema
-```sql
-CREATE TABLE bookmarks (
-  id INTEGER PRIMARY KEY,
-  url TEXT NOT NULL,
-  title TEXT NOT NULL,
-  created_at DATETIME DEFAULT CURRENT_TIMESTAMP,
-  updated_at DATETIME DEFAULT CURRENT_TIMESTAMP
-);
-
-CREATE TABLE tags (
-  id INTEGER PRIMARY KEY,
-  name TEXT UNIQUE NOT NULL
-);
-
-CREATE TABLE bookmark_tags (
-  bookmark_id INTEGER REFERENCES bookmarks(id),
-  tag_id INTEGER REFERENCES tags(id),
-  PRIMARY KEY (bookmark_id, tag_id)
-);
-```
-
-## Requirements
-- TypeScript throughout
-- Input validation (frontend + backend)
-- Error handling with user feedback
-- Loading states
-- Empty states
-- Responsive design
-
-## Testing
-- Backend: Jest + supertest for API tests
-- Frontend: Basic component tests (optional)
-- E2E: Manual testing checklist
-
-## Out of Scope
-- User authentication
-- Import/export
-- Browser extension
-- Cloud deployment
-- Real-time sync
-
-## Success Criteria
-- All CRUD operations work
-- Search and filter work
-- No console errors
-- Tests pass
-- Code review passes (all 3 reviewers)
-
----
-
-**Purpose:** Comprehensive test of Loki Mode's full capabilities including frontend, backend, database, and code review agents. Expect ~30-60 minutes for full execution.

package/skills/loki-mode/examples/simple-todo-app.md

@@ -1,60 +0,0 @@
-# PRD: Simple Todo App
-
-## Overview
-A minimal todo application for testing Loki Mode with a simple, well-defined scope.
-
-## Target Users
-Individual users who want a simple way to track tasks.
-
-## Features
-
-### MVP Features
-1. **Add Todo** - Users can add a new todo item with a title
-2. **View Todos** - Display list of all todos
-3. **Complete Todo** - Mark a todo as done
-4. **Delete Todo** - Remove a todo from the list
-
-### Tech Stack (Suggested)
-- Frontend: React + TypeScript
-- Backend: Node.js + Express
-- Database: SQLite (local file)
-- No deployment (local testing only)
-
-## Acceptance Criteria
-
-### Add Todo
-- [ ] Input field for todo title
-- [ ] Submit button
-- [ ] New todo appears in list
-- [ ] Input clears after submit
-
-### View Todos
-- [ ] Shows all todos in a list
-- [ ] Shows completion status
-- [ ] Empty state when no todos
-
-### Complete Todo
-- [ ] Checkbox or button to mark complete
-- [ ] Visual indicator for completed items
-- [ ] Persists after refresh
-
-### Delete Todo
-- [ ] Delete button on each todo
-- [ ] Confirmation before delete
-- [ ] Removes from list and database
-
-## Out of Scope
-- User authentication
-- Due dates
-- Categories/tags
-- Mobile app
-- Cloud deployment
-
-## Success Metrics
-- All features functional
-- Tests passing
-- No console errors
-
----
-
-**Purpose:** This PRD is intentionally simple to allow quick testing of Loki Mode's core functionality without waiting for complex builds or deployments.

package/skills/loki-mode/examples/static-landing-page.md

@@ -1,73 +0,0 @@
-# PRD: Static Landing Page
-
-## Overview
-A simple static landing page for a fictional SaaS product. Tests Loki Mode's frontend and marketing agent capabilities.
-
-## Target Users
-Marketing teams needing a quick landing page.
-
-## Page Sections
-
-### Hero Section
-- Headline: "Supercharge Your Workflow"
-- Subheadline: "The all-in-one tool for modern teams"
-- Primary CTA: "Get Started Free"
-- Secondary CTA: "Watch Demo"
-- Hero image placeholder
-
-### Features Section (3 features)
-1. **Fast Setup** - "Get started in minutes, not days"
-2. **Team Collaboration** - "Work together seamlessly"
-3. **Analytics** - "Track what matters"
-
-### Social Proof
-- 3 testimonial cards with placeholder content
-- "Trusted by 10,000+ teams"
-
-### Pricing Section
-- Free tier: $0/month
-- Pro tier: $29/month
-- Enterprise: Contact us
-
-### FAQ Section
-- 4 common questions with answers
-
-### Footer
-- Links: About, Blog, Careers, Contact
-- Social icons: Twitter, LinkedIn, GitHub
-- Copyright notice
-
-## Tech Stack
-- HTML5
-- CSS3 (no framework, or Tailwind CSS)
-- Minimal JavaScript (for FAQ accordion)
-- No build step required
-
-## Requirements
-- Responsive design (mobile + desktop)
-- Semantic HTML
-- Accessible (WCAG 2.1 AA basics)
-- Fast load time (< 2s)
-- No external dependencies (except fonts)
-
-## Assets
-- Use placeholder images (placeholder.com or similar)
-- Use system fonts or Google Fonts
-- Use emoji for icons if needed
-
-## Out of Scope
-- Backend/API
-- Form submission handling
-- Analytics tracking
-- A/B testing
-- Deployment
-
-## Deliverables
-1. `index.html` - Main page
-2. `styles.css` - Stylesheet
-3. `script.js` - Minimal JS (optional)
-4. `README.md` - How to view locally
-
----
-
-**Purpose:** Tests frontend agent, marketing agent (copy), and design patterns without backend complexity.

package/skills/loki-mode/examples/todo-app-generated/.loki/CONTINUITY.md

@@ -1,59 +0,0 @@
-# Loki Mode Working Memory
-Last Updated: 2026-01-02T23:55:00Z
-Current Phase: completed
-Current Iteration: Final
-
-## Active Goal
-Simple Todo App - COMPLETED ✅
-
-## Current Task
-- ID: ALL TASKS COMPLETED
-- Description: All 18 tasks successfully executed
-- Status: completed
-- Completion Time: ~15 minutes (with Haiku parallelization)
-
-## Just Completed
-ALL TASKS (001-018):
-- task-001: Project structure ✅
-- task-002: Backend initialization ✅
-- task-003: Frontend initialization ✅
-- task-004: Database setup ✅
-- task-005-008: API endpoints (parallel execution) ✅
-- task-009: API client ✅
-- task-010: useTodos hook ✅
-- task-011-012: TodoForm & TodoItem (parallel) ✅
-- task-013-015: TodoList, EmptyState, ConfirmDialog ✅
-- task-016: App assembly ✅
-- task-017: CSS styling ✅
-- task-018: E2E testing ✅
-
-## Performance Metrics
-- Total Tasks: 18
-- Completed: 18 (100%)
-- Failed: 0
-- Haiku Agents Used: 14
-- Sonnet Agents Used: 0
-- Opus Agents Used: 1 (architecture planning)
-- Parallel Executions: 3 batches (tasks 002-003, 005-008, 011-012)
-- Estimated Time Saved: 8x faster with parallelization
-
-## Active Blockers
-- (none)
-
-## Key Decisions This Session
-- Using Simple Todo App PRD for test
-- Local-only deployment (no cloud)
-- Tech Stack: React + TypeScript (frontend), Node.js + Express (backend), SQLite (database)
-
-## Working Context
-System starting fresh. Testing Loki Mode v2.16.0 with example PRD.
-PRD Requirements:
-- Add Todo (title input, submit button)
-- View Todos (list display, completion status)
-- Complete Todo (checkbox/button, visual indicator)
-- Delete Todo (delete button with confirmation)
-- No auth, no deployment, local testing only
-
-## Files Currently Being Modified
-- .loki/CONTINUITY.md: initialization
-- .loki/state/orchestrator.json: system state

package/skills/loki-mode/examples/todo-app-generated/.loki/queue/completed.json

@@ -1 +0,0 @@
-{"tasks":[]}

package/skills/loki-mode/examples/todo-app-generated/.loki/queue/dead-letter.json

@@ -1 +0,0 @@
-{"tasks":[]}

package/skills/loki-mode/examples/todo-app-generated/.loki/queue/failed.json

@@ -1 +0,0 @@
-{"tasks":[]}

package/skills/loki-mode/examples/todo-app-generated/.loki/queue/in-progress.json

@@ -1 +0,0 @@
-{"tasks":[]}