Skip to content

on_boarding.md

Latest commit

 

History

History
89 lines (53 loc) · 3.98 KB

on_boarding.md

File metadata and controls

89 lines (53 loc) · 3.98 KB
Raw
graph LR
    Arcade_Core["Arcade Core"]
    Serving_Layer["Serving Layer"]
    Developer_CLI["Developer CLI"]
    Toolkits["Toolkits"]
    Framework_Adapters["Framework Adapters"]
    Evaluation_Framework["Evaluation Framework"]
    Developer_CLI -- "Manages Deployments of" --> Toolkits
    Developer_CLI -- "Deploys to" --> Serving_Layer
    Toolkits -- "Implements" --> Arcade_Core
    Framework_Adapters -- "Consumes Tools from" --> Serving_Layer
    Serving_Layer -- "Executes Tools using" --> Arcade_Core
    Evaluation_Framework -- "Validates" --> Toolkits
    Evaluation_Framework -- "Tests using" --> Arcade_Core
    click Arcade_Core href "https://github.com/CodeBoarding/GeneratedOnBoardings/blob/main/arcade-ai/Arcade_Core.md" "Details"
    click Serving_Layer href "https://github.com/CodeBoarding/GeneratedOnBoardings/blob/main/arcade-ai/Serving_Layer.md" "Details"
    click Evaluation_Framework href "https://github.com/CodeBoarding/GeneratedOnBoardings/blob/main/arcade-ai/Evaluation_Framework.md" "Details"
Loading

CodeBoardingDemoContact

Details

An analysis of the Arcade AI platform's architecture, based on the provided context and Control Flow Graph (CFG) data, reveals a modular and scalable system designed for developing and serving AI agent tools. The platform's components are clearly delineated, with arcade-core serving as the central hub providing foundational abstractions and execution logic. The primary workflow involves developers creating Toolkits using these core abstractions, which are then deployed to the Serving Layer via the Developer CLI. Once live, these tools are consumed by AI agents, a process simplified by Framework Adapters. The Evaluation Framework supports this lifecycle by enabling robust testing and validation of the toolkits.

Arcade Core [Expand]

The foundational engine providing core abstractions, data schemas (Toolkit), and the ToolExecutor for running tools. It is the central dependency for all other platform components.

Related Classes/Methods:

  • arcade_core.catalog
  • arcade_core.executor
  • arcade_core.toolkit

Serving Layer [Expand]

Exposes toolkits over a network via a worker-based system (FastAPIWorker). It handles incoming requests from AI agents and orchestrates tool execution using Arcade Core.

Related Classes/Methods:

  • arcade_serve.fastapi.worker
  • arcade_serve.mcp.server

Developer CLI

The primary interface for developers. It provides commands to package, deploy, and manage toolkits on the Serving Layer, and handles developer authentication.

Related Classes/Methods:

  • arcade_cli.main
  • arcade_cli.deployment.packages

Toolkits

User-created packages containing the business logic for a set of tools. They are built using the abstractions provided by Arcade Core.

Related Classes/Methods:

  • toolkits.postgres.arcade_postgres.database_engine

Framework Adapters

A set of modules that act as clients to the Serving Layer, enabling seamless integration and consumption of Arcade tools within popular AI agent frameworks like LangChain.

Related Classes/Methods:

  • contrib.langchain.langchain_arcade.manager
  • contrib.crewai.crewai_arcade.manager

Evaluation Framework [Expand]

Provides the necessary tools (EvalSuite, Critic) to test and validate the functionality and performance of toolkits, interacting directly with the core execution logic.

Related Classes/Methods:

  • arcade_evals.eval
  • arcade_evals.critic

FAQ