EventureAI publishes a Model Context Protocol server at .well-known/mcp.json — the open standard for AI agent discovery. Any MCP-compatible client can discover our capabilities, authenticate, and begin operating against your business data in seconds. This is how AI stops being a chatbot and starts being infrastructure.
Model Context Protocol defines a standard way for AI agents to discover what a server can do. By publishing a .well-known/mcp.json file at the root of our domain, any MCP-compatible AI client — Claude, GPT, custom agents, automation frameworks — can automatically discover EventureAI's capabilities, endpoints, authentication requirements, and available tools. No manual configuration. No reading documentation and copy-pasting URLs. The agent reads the discovery file and knows exactly how to connect. This is the same pattern that made OAuth, WebFinger, and SSL certificate discovery work at internet scale. One standard file. Universal machine readability. Instant integration.
Our MCP server exposes three categories of capabilities: Tools — executable actions that AI agents can invoke against your business data. Query your ledger, check inventory, generate reports, post journal entries, dispatch technicians, send notifications. Every tool is authenticated, scoped to your tenant, and logged to the audit trail. Resources — read-only data that AI agents can access. Tenant profiles, media assets, page arrays, route tables, integration status, SEO page inventories. The resources give agents context about your business without requiring them to execute queries. Prompts — pre-built interaction templates that agents can use to perform common business operations. Onboard a new customer, process a service call, reconcile a day's transactions, generate an operations report. The prompts encode business logic so agents don't have to reinvent it.
Visit eventureai.com/.well-known/mcp.json to see the live discovery file. It includes: the server name and description, the base URL for all API interactions, the authentication method (bearer token with waas_sk_ prefix), the available endpoint categories (CLI, health, builder), and contact information for requesting access. The file is served with CORS headers so browser-based agents can discover it. It is cached for one hour so discovery is fast. And it is the single source of truth for how external systems connect to EventureAI.
Here is how a typical AI agent integration works with our MCP server: Step 1 — The agent fetches.well-known/mcp.json and parses our capabilities. Step 2 — The agent authenticates with a bearer token issued through your admin dashboard. Step 3 — The agent queries the CLI endpoints to understand your business state — tenants, routes, health, open prompts. Step 4 — The agent takes actions based on what it discovers — dispatching technicians, generating SEO pages, posting accounting entries, sending customer notifications. Step 5 — Every action is logged, hashed, and provable on blockchain. The agent operates under the same audit standards as a human operator. This is not theoretical.
Underneath the MCP layer sits our proprietary Array protocol — the encoding format that compresses business operations into compact, executable arrays. The format: [type, tenant, instruction, dependencies, dimensions, layout]. Type 0 is human input. Type 1 is SQL. Type 2 is JavaScript. Type 3 is media. One array row in the database can drive 50,000 devices. The device is the runtime. The LLM is the dictionary. The human is the seed. MCP is how agents discover and connect. Arrays are how the system thinks, stores, and executes. Together they form a complete machine-to-machine business operating system — stateless, provable, and infinitely scalable.
Discover our MCP server at eventureai.com/.well-known/mcp.json or request API access.