skills.rest
nodefony
nodefonyOFFICIAL
Explore all skills by this author →
WebsiteGitHub

nodefony-roadmap

Provides design context and invariants for Nodefony's agentic AI layer roadmap.

Updated Dec 19, 2023
Search Tags:#llm#rag#roadmap#agentic ai#websocket#nodefony#design invariants
One-click install
npx skills add https://github.com/nodefony/nodefony-core --skill nodefony-roadmap-nodefony
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: nodefony-roadmap
Source: https://github.com/nodefony/nodefony-core/tree/main/.claude/skills/nodefony-roadmap
Command: npx skills add https://github.com/nodefony/nodefony-core --skill nodefony-roadmap-nodefony

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Working on Nodefony's AI agentic layer (Phase 12) without knowing its design invariants, module states, and phase dependencies risks breaking work-in-progress modules or violating frozen architectural decisions. ## Core Features & Use Cases - Phase 12 AI Layer Context: Documents the planned modules (@nodefony/llm, vector, rag, memory, agent, agent-guard) with their sub-phases and real implementation status. - Design Invariants: Enforces seven non-negotiable rules including generic modules, dependency injection, native streaming, human validation, sovereign mode, AI Act compliance, and WebSocket as LLM transport. - Delivered Phase Pointers: Redirects conventions from completed phases (Studio P10, Realtime P13, Vite builder P14) to the skills and docs that now own them. - Use Case: Before designing an interface consumed by an AI agent, consult this Skill to apply the invariants and respect the rule against starting AI module sessions before Phase 6 completes. ## Quick Start Ask the agent to review the design invariants and module status before starting work on any @nodefony AI module or admin API convention.

Frequently Asked Questions about nodefony-roadmap

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
What modules make up Nodefony's agentic AI layer?▼

The Phase 12 AI layer includes @nodefony/llm for multi-LLM support, @nodefony/vector for vector adapters, @nodefony/rag for RAG pipelines, @nodefony/memory for agent memory, @nodefony/agent for orchestration, and @nodefony/agent-guard for governance, zones, PII, audit, and cost control.

How does Nodefony handle LLM streaming over WebSocket?▼

Nodefony uses its native WebSocket pipeline from @nodefony/http as the LLM transport, with server-side AsyncGenerator<string> streaming. Since the RPC protocol has no fragment streaming, progress uses a work-plus-channel pattern where the action acknowledges and progress arrives on a subscribed channel.

Can I start working on Nodefony AI modules before Phase 6?▼

No, sessions on AI modules should not start before Phase 6 completes unless explicitly requested, because the design is not frozen and a refactor is planned in P12.1. Modules consumed by AI should still design for AI usage with extensible interfaces and async iterators.

What is the difference between @nodefony/frontend and nodefony/client?▼

@nodefony/frontend is the Vite builder that registers module frontends, while nodefony/client is the browser library shipped as a subpath of the core package. A separate @nodefony/client package must not be recreated per the frozen subpaths decision.

What are the limitations of Nodefony's RPC protocol for streaming?▼

The RPC protocol has no streaming support; an action returns a single value with no fragment frames. True RPC streaming with cancellation, mid-stream errors, and flow control is deferred to Phase 12 design with a real first consumer.

Related Skills

Explore More Software Engineering Skills→
tonehqtonehq
11

websocket

Establish real-time WebSocket communication for tone's transport layer.

Official
Intermediate
adamrdrewadamrdrew
7

describe-laws-and-style

Documents invariants and conventions for Ushabti-based projects across Phases and refactors.

Community
Basic
Binh230199Binh230199
1

android-architecture

Design and review Android app architecture for MVVM or MVI feature modules.

Community
Advanced
simplysmartaisimplysmartai

langchain-architecture

Design LLM applications with LangChain agents, chains, memory, and tool integration.

Community
Advanced