How agenticsorg/edge-agents works
Main Components
– Agent Functions (e.g., agent_alpha, agent_beta, agent_stream, agent_websocket, agentic_inbox_agent)
– Management Functions (e.g., agent-manager, edge_deployment)
– Communication Functions (e.g., resend, send-contact-notification)
– Integration Functions (e.g., github-api, mcp-server)
– Utility Functions (e.g., stripe_check-subscription-status, feedback, test-function)
• MCP Server and Agent Orchestration: The “scripts/agentic-mcp” package appears to implement the Model Context Protocol server. Represent its role in coordinating multi-agent workflows, performing web research, summarization, context management, and offering additional tools.
• Agent Implementations: There are multiple agent projects throughout the file tree (e.g., agents/*, scripts/agents, scripts/openai-agent, gemini-rotator, gemini-tumbler, sparc2). These components show examples of agent clients, orchestration, and even testing frameworks.
• Supporting Tools & Libraries: In addition to the agents and MCP server, include components that provide utility functions for:
– Database operations (database integration with Supabase/PostgreSQL)
– Real-time communication (using Supabase’s channels)
– Secure secrets management
– Logging/tracing and debugging (as seen by tracing modules within agents and extensive tests)
• External Services and Infrastructure: Highlight the use of:
– Supabase (for edge functions, real-time capabilities, and database interactions)
– Deno (the runtime powering many of these functions)
– Third-party integrations (like OpenAI APIs, GitHub APIs, Stripe, Resend, Calendly, etc.)
Relationships and Interactions
• Internal agent communication and orchestration. For example, an incoming request might be processed by an agent function that uses the MCP server to delegate tasks to specialized agents.
• How the management functions allow deployment and dynamic workflow management between agents.
– Between client calls and edge functions (low latency, secure access)
– Between the MCP server and individual agents (orchestration and context management)
– Between edge functions and external services (database queries, authentication, and external API calls)
Architectural Patterns and Design Principles
• Microservice or modular design: Each function or agent is an independent component that can be developed, tested, and deployed individually.
• Separation of concerns: Different folders are dedicated to agents, management, integration, and utility functions.
• Event-driven and real-time: Emphasize how the platform leverages Supabase’s real-time channels and edge function architecture.
• Security-first: Illustrate the use of Deno with explicit permissions and secure handling of environment variables and secrets.
• Scalability: Note the dynamic scaling aspect tied to serverless deployment at the network edge.