Skip to content

data-flow.md

Latest commit

 

History

History
90 lines (78 loc) · 4.23 KB

data-flow.md

File metadata and controls

90 lines (78 loc) · 4.23 KB

Data Flow

Overview

This document describes the data flow within the Agentic DevOps framework, illustrating how data is processed and passed between different components during typical operations.

Data Flow Diagrams

1. Command Execution Data Flow

sequenceDiagram
    participant User
    participant CLI (cli.py)
    participant ArgumentParser
    participant Config Module
    participant Credential Manager
    participant EC2 Service
    participant GitHub Service
    participant AWS API
    participant GitHub API

    User->>CLI: Executes CLI command (e.g., `run_cli.py ec2 list-instances`)
    CLI->>ArgumentParser: Parse command and arguments (using argparse)
    ArgumentParser->>CLI: Returns parsed arguments
    CLI->>Config Module: Get configuration (e.g., AWS region)
    Config Module->>CLI: Returns configuration data
    CLI->>Credential Manager: Get credentials (e.g., AWS credentials)
    Credential Manager->>CLI: Returns credentials
    alt EC2 Command
        CLI->>EC2 Service: Call EC2 service function (e.g., `list_instances`) with arguments and context
        EC2 Service->>AWS API: Call AWS API (e.g., `ec2.describe_instances`) with credentials
        AWS API->>EC2 Service: Returns AWS API response data (e.g., instance details)
        EC2 Service->>CLI: Returns processed instance data
    else GitHub Command
        CLI->>GitHub Service: Call GitHub service function (e.g., `list_repositories`) with arguments and context
        GitHub Service->>GitHub API: Call GitHub API (e.g., `GET /user/repos`) with credentials
        GitHub API->>GitHub Service: Returns GitHub API response data (e.g., repository list)
        GitHub Service->>CLI: Returns processed repository data
    end
    CLI->>User: Display formatted output to user (e.g., table or JSON)
Loading

2. Agent Workflow Data Flow

sequenceDiagram
    participant Agent
    participant Runner
    participant Tool Registry
    participant Tool (e.g., list_ec2_instances)
    participant RunContext
    participant Guardrail System
    participant External Service (AWS, GitHub)

    User->>Runner: Start agent workflow with task input
    Runner->>Agent: Initialize agent with task and context
    Agent->>Tool Registry: Select appropriate tool based on task
    Agent->>RunContext: Access context information (e.g., region, credentials)
    Runner->>Guardrail System: Input Guardrail Check (e.g., security_guardrail)
    Guardrail System->>Runner: Returns guardrail check result
    alt Guardrail Passed
        Runner->>Tool: Execute tool (e.g., `list_ec2_instances`) with context
        Tool->>External Service: Interact with external service API (AWS, GitHub)
        External Service->>Tool: Returns API response data
        Tool->>Runner: Returns processed tool output
        Runner->>Agent: Provide tool output to agent
        Agent->>Runner: Generate final output or next action
        Runner->>Guardrail System: Output Guardrail Check (e.g., sensitive_info_guardrail)
        Guardrail System->>Runner: Returns output guardrail check result
        alt Guardrail Passed
            Runner->>User: Return final result to user
        else Guardrail Failed
            Runner->>User: Return guardrail violation message
        end
    else Guardrail Failed
        Runner->>User: Return guardrail violation message
    end
Loading

Data Entities

  • Configuration Data: Loaded from YAML/JSON files or environment variables. Managed by config.py. Includes settings for AWS, GitHub, logging, etc.
  • Credential Data: AWS credentials (access key, secret key, token, region, profile) and GitHub token, managed by credentials.py and CredentialManager.
  • Context Data: DevOpsContext object containing user_id, aws_region, github_org, environment, and metadata, managed by context.py.
  • Request Data: User commands from CLI or task inputs for agents, typically strings or structured data.
  • Response Data: JSON responses from AWS and GitHub APIs, processed by service modules and tools.
  • Output Data: Formatted CLI output (JSON, table) or agent workflow results, which can be text, structured data, or actions.

This document provides a detailed overview of the data flow within the Agentic DevOps framework for both command execution and agent workflows.