Skip to content

augmented-llm.md

Latest commit

 

History

History
120 lines (83 loc) · 4.48 KB

augmented-llm.md

File metadata and controls

120 lines (83 loc) · 4.48 KB
Raw

🏠 Home › Foundations › 🦄 Augmented LLM

← Foundations ━━━━━━━━━━━━━━━━━━●━━━━━━━━━━━━━━━━━━ Workflows →

🦄 The Augmented LLM

TL;DR: The foundation of ALL agentic systems — an LLM enhanced with retrieval, tools, and memory. Every workflow builds on this.

Diagram

%%{init: {'theme': 'base', 'themeVariables': {'lineColor': '#64748b'}}}%%
flowchart TB
    classDef retrieval fill:#3b82f6,stroke:#2563eb,stroke-width:2px,color:#ffffff
    classDef tools fill:#8b5cf6,stroke:#7c3aed,stroke-width:2px,color:#ffffff
    classDef memory fill:#06b6d4,stroke:#0891b2,stroke-width:2px,color:#ffffff
    classDef llm fill:#f59e0b,stroke:#d97706,stroke-width:2px,color:#ffffff

    R[("Retrieval<br/>RAG, search")]:::retrieval
    T{{"Tools<br/>MCP, Bash"}}:::tools
    M[/"Memory<br/>Context"/]:::memory

    LLM(["LLM<br/>Generates • Selects • Decides"]):::llm

    R --> LLM
    T --> LLM
    M --> LLM
Loading

Key Components

Component Description Examples
Retrieval Access to external knowledge RAG, docs, search
Tools Ability to take actions MCP, Bash, API calls
Memory Context persistence Conversation history, state
LLM Core reasoning engine Claude, GPT, etc.

Key Insight

Focus on tailoring capabilities to your specific use case and ensuring they provide an easy, well-documented interface for the LLM.

All workflows assume each LLM call has access to these augmented capabilities.

Key Distinction

┌─────────────────────────────────────────────────────────────────────────────┐
│                    🦄 BUILDING BLOCK = AUGMENTED LLM                        │
│                       (foundation for ALL workflows)                        │
├───────────────┬───────────────┬───────────────┬─────────────────────────────┤
│   Retrieval   │    Tools      │    Memory     │            LLM              │
│   (RAG/docs)  │   (actions)   │   (context)   │           (core)            │
└───────────────┴───────────────┴───────────────┴─────────────────────────────┘
  • 🦄 Building Block = Augmented LLM (Anthropic's foundation concept)
  • Components = Claude Code abstractions (🐦 Subagent, 🦴 Slash Command, 📚 Skill, 🪝 Hook)
  • Layers = Architectural organization (User → Main Agent → Delegation → Execution → State)

Terminology

Symbol Term Description
🐔 Main Agent Claude Code orchestrator (the hen that coordinates)
🐦 Subagent Delegated worker spawned via Task (the bird)
🪺 Spawn (Task) Action to create 🐦 subagents (via Task built-in tool)
📚 Skill Loaded knowledge that enhances 🐔 capabilities
🚧 Gate Checkpoint that validates output before proceeding

Hierarchy

%%{init: {'theme': 'base', 'themeVariables': {'lineColor': '#64748b'}}}%%
flowchart LR
    classDef user fill:#6366f1,stroke:#4f46e5,stroke-width:2px,color:#ffffff
    classDef main fill:#f59e0b,stroke:#d97706,stroke-width:2px,color:#ffffff
    classDef sub fill:#ec4899,stroke:#db2777,stroke-width:2px,color:#ffffff
    classDef blocked fill:#ef4444,stroke:#dc2626,stroke-width:2px,color:#ffffff

    U1["🙋‍♀️📥 User"]:::user
    MA["🐔 Main Agent"]:::main
    SA["🐦 Subagent"]:::sub
    U2["💁‍♀️📤 User"]:::user

    U1 -->|request| MA
    MA -->|"🪺 spawn"| SA
    SA -->|result| MA
    MA -->|response| U2

    SA x-.-x|"❌ cannot spawn"| SA2["🐦 Subagent"]:::blocked
Loading

Critical Rule: 🐦 Subagents CANNOT spawn other 🐦 subagents (flat hierarchy)

← Foundations ━━━━━━━━━━━━━━━━━━●━━━━━━━━━━━━━━━━━━ Workflows →