Skip to content

Latest commit

 

History

History
56 lines (52 loc) · 6.51 KB

File metadata and controls

56 lines (52 loc) · 6.51 KB

Implementation Status

This page tracks the implementation status of various Agent Development Kit (ADK) features within AgentLabUI and related platforms.

Legend:

  • ✅ : Implemented / Natively Supported by the UI
  • 🚧 : Not yet implemented in the UI / Support is indirect or requires manual ADK coding outside the UI
  • ❓ : Status Unknown / To Be Determined
  • 🙈 : Feature not available in this framework
  • ❌ : Will not implement / Not in Roadmap
Feature Category Feature Name AgentLabUI (via ADK) LlamaStack AWS Bedrock
Agent Definition & Types
LlmAgent (reasoning agent)
Workflow Agents (Sequential, Parallel, Loop)
Custom Agents (BaseAgent inheritance) 🚧
Multi-Agent Systems (hierarchy, delegation)
Tooling
Function Tools (custom, via Gofannon)
Agent-as-a-Tool (via Child Agents for orchestration)
Long Running Function Tools 🚧
Built-in: Google Search
Built-in: Code Execution
Built-in: Vertex AI Search
Third-Party Tools (LangChain, CrewAI via Gofannon)
OpenAPI Toolset Integration 🚧
MCP Toolset (ADK as client for external MCP)
Exposing ADK tools via custom MCP Server 🚧
Google Cloud Tools (API Hub, App Int., DB Toolbox) 🚧
Tool Authentication Support (via Gofannon tool config)
Runtime & Orchestration
Runner & Event Loop ✅ (Implicit)
Event System & History Logging
Session Management (Session, State) ✅ (Implicit)
State Scoping (user:, app:, temp:) 🚧
MemoryService (Long-term knowledge) 🚧
ArtifactService (Binary data management) 🚧
LLM Integration
Model Agnostic (ADK core) ✅ (ADK Core)
Gemini Integration (Vertex AI)
LiteLLM Integration (OpenAI, Anthropic, etc.) 🚧
Development & Deployment
ADK Web UI (official adk web) 🚧 (AgentLabUI is separate)
ADK CLI (adk run, etc.) 🚧 (ADK tool)
Deployment to Vertex AI Agent Engine
Deployment to Cloud Run / GKE 🚧
Advanced Features
Callbacks (before/after agent, model, tool) 🚧
Streaming (Text, Audio, Video - for deployed agents) 🚧 (Text-based for AgentLabUI runner)
Streaming Tools 🚧
Evaluation Framework (ADK native) 🚧
Safety Guardrails (via Callbacks, Tool Design) 🚧
Sandboxed Code Execution (via ADK + Vertex) ✅ (If ADK uses Vertex Code Exec)
Tracing / Observability (Basic Event Log)