Product Architecture — Unified Agent Platform

Documents the blueprint for building a multi-protocol AI assistant system with graded user approval controls.

Updated Apr 28, 2025
One-click install
npx skills add https://github.com/tim-o-private/llm-agent --skill product-architecture-unified-agent-platform
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: Product Architecture — Unified Agent Platform
Source: https://github.com/tim-o-private/llm-agent/tree/main/.claude/skills/product-architecture
Command: npx skills add https://github.com/tim-o-private/llm-agent --skill product-architecture-unified-agent-platform

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

This Skill provides a comprehensive blueprint for building a unified AI agent platform, ensuring consistency and safety across all interaction channels.

Core Features & Use Cases

  • Unified Architecture: Enables agents to operate seamlessly across web, Telegram, and scheduled jobs, sharing state and context.
  • Tiered Approval System: Enforces safety by classifying tool access into auto-approve, requires approval, and user-configurable tiers.
  • Cross-Cutting Primitives: Offers reusable components like a jobs queue, notification service, and pending actions system to streamline development.
  • Use Case: Develop a new channel integration (e.g., Slack) by leveraging existing primitives for agent execution, user notification, and action approval, ensuring a consistent user experience.

Quick Start

Follow the recipe to add a new channel by creating a channel-specific handler and integrating with the shared agent execution pipeline.

Frequently Asked Questions about Product Architecture — Unified Agent Platform

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

FAQPage Schema
What is a unified agent platform for multi-channel AI?

A unified agent platform enables AI agents to operate seamlessly across web, Telegram, and scheduled jobs while sharing state, context, and cross-cutting primitives like a jobs queue and notification service.

How do I add a new channel to a multi-channel AI agent platform?

To add a new channel, create a channel-specific handler and integrate it with the shared agent execution pipeline, leveraging existing primitives for user notification and action approval to maintain a consistent experience.

How does a tiered approval system work for AI agent tool usage?

A tiered approval system enforces safety by classifying tool access into auto-approve, requires approval, and user-configurable tiers, ensuring controlled execution of agent actions across all interaction channels.

Can I use a shared jobs queue and notification service for scheduled AI jobs?

Yes, the unified architecture provides reusable cross-cutting primitives, including a shared jobs queue, notification service, and pending actions system, to streamline scheduled job execution and cross-channel state management.

What's the best way to maintain consistent state across web and Telegram AI agents?

Implementing a unified agent platform with shared architectural primitives allows agents to share state and context seamlessly across web, Telegram, and other channels, ensuring consistent behavior and user experience.