Capability Decomposition

14 Engineering Patterns

Click any discovery for full technical analysis, code examples, architecture diagrams, and LightHope ecosystem mapping.

Priority

Category

Showing 24 of 24 discoveries

KAIROS — Always-On Daemon

A persistent background agent that never sleeps

autoDream — Memory Consolidation

AI that dreams — reorganizing memory while you sleep

ULTRAPLAN — Cloud Deep Planning

30 minutes of autonomous deep thinking on cloud Opus

Coordinator Mode — Multi-Agent Orchestration

One Claude orchestrating a team of parallel worker agents

BUDDY — Companion Pet System

A Tamagotchi-style terminal pet with 18 species and gacha mechanics

44 Feature Flags — Hidden Roadmap

A complete product roadmap hiding behind compile-time gates

Undercover Mode — Stealth Attribution

Auto-strips AI identity when contributing to public repos

Anti-Distillation — Training Data Poisoning

Fake tools injected to corrupt competitor training data

Context Entropy Solution — 5 Compaction Strategies

How to prevent AI from going crazy in long sessions

Internal Model Codenames

Tengu, Capybara, Fennec, Numbat — and what comes next

Advisor Agent — Session Overseer

A second AI watching the first AI work

Tool Architecture — 40+ Plugin System

Every capability is a discrete, permission-gated tool

User Frustration Detection

Regex-based sentiment analysis — 'the world's most expensive company using regex'

UDS Inbox — Inter-Agent Communication

Unix domain sockets for agent-to-agent messaging on the same machine

Extended Thinking — Deep Reasoning Budget

Allocating compute budget for chain-of-thought before responding

Managed Agents API — Server-Side Agent Lifecycle

Persistent server-side agents with sessions and memory

MCP — Model Context Protocol Architecture

Universal standard for connecting AI to external tools and data

Prompt Caching — Token Cost Optimization

Cache static prompt sections to cut costs by 90%

Computer Use — GUI Automation via Vision

AI that sees and clicks — visual interface automation

Files API — Persistent Content Storage

Upload files once, reference across multiple conversations

Agent SDK — Programmatic Agent Loop

Build custom agents in code — the loop is yours to control

Streaming Tool Use — Real-Time Agent Actions

Stream agent reasoning and tool calls as they happen

Ambient Agents — Background Autonomous Execution

AI agents that work while you sleep

Hooks System — Standards as Automated Checkpoints

Every standard becomes an executable checkpoint