AI Agents Integration Overview

This document summarizes the scaffolding implemented for the Syntax & Sips AI workflows.

Components

Agents: Coordinator, Research, Writing, Editing, Optimization stubs live in src/agents/ with shared types and retry utilities.
MCP Client Layer: Located in src/lib/mcp/ providing typed helpers for research, blog, SEO, and storage servers.
Services: Workflow management utilities in src/services/ai/ handle Supabase persistence and event broadcasting.
API Routes: Endpoints under src/app/api/ai/ expose workflow management, draft access, and tool execution with SSE streaming support.
Dashboard UI: The dashboard route src/app/(dashboard)/ai renders launch, timeline, draft, and SEO insight panels.
MCP Servers: Four lightweight servers are defined under apps/mcp/* for research, blog, SEO, and storage operations.

Install dependencies at the project root with npm install.
Each MCP server has its own package.json; install dependencies with npm install inside the server directory before running.

Start the research server as an example:

cd apps/mcp/research
npm install
npm run dev

Configure environment variables for API routes and client helpers:
- MCP_RESEARCH_URL
- MCP_BLOG_URL
- MCP_SEO_URL
- MCP_STORAGE_URL
- Optional auth tokens such as MCP_RESEARCH_TOKEN.
Run npm run dev at the project root to access the dashboard at /ai (under the dashboard segment).

Run Supabase migrations to create AI workflow tables:

supabase db push

The new migration adds workflow state, events, research notes, and SEO metadata columns on posts.

Basic coordinator and retry unit coverage is provided via Playwright test harness in tests/ai-agents.spec.ts.

Configure Vercel Analytics or an APM provider to watch dashboard performance (npm run build ensures instrumentation hooks run).
Aggregate MCP server logs via pino transports (e.g., ship to Datadog or Logflare) to track error rates and latency.
Add heartbeat checks for each MCP server endpoint (/mcp) using your uptime monitor of choice.

Replace in-memory storage used by the MCP blog and storage servers with Supabase integrations.
Expand agent implementations to call MCP clients and emit workflow events.
Add comprehensive end-to-end scenarios covering dashboard streaming behaviour.