Retool

OKF knowledge bundle (/okf/index.md)

No OKF bundle. Google's Open Knowledge Format (June 2026) lets AI agents navigate your business knowledge as a graph of linked markdown files. Less than 0.1% of sites have this.

Fix: Create an OKF bundle starting with /okf/index.md. Use Faro's OKF Generator to build a complete bundle automatically.

/pricing.md (machine-readable pricing)

No /pricing.md. AI procurement agents that evaluate and compare software prefer plain-text pricing they can read directly , no JavaScript, no tables to parse. Buffer, Resend, WorkOS, and Supabase all publish /pricing.md.

Fix: Create a plain-text Markdown file at /pricing.md. It should include: - Plan names and exact prices (e.g. "$9/month") - What's included in each plan (bullet list) - Annual pricing if available - A link to your full pricing page Serve it at /pricing.md with content-type text/markdown or text/plain. In Next.js, place the file in /public/pricing.md , Next.js serves public/ files automatically. Example: https://byfaro.ai/pricing.md

/agents.json (agent capability manifest)

No /agents.json. This file tells AI agents what actions they can take, what APIs you expose, and how to interact with your business.

Fix: Create /agents.json at your root: { "name": "Your Company", "description": "What you do", "url": "https://yourdomain.com", "capabilities": ["Browse products", "Check pricing", "Start trial"], "api": { "docs": "https://yourdomain.com/docs" }, "pricing_url": "https://yourdomain.com/pricing", "auth": ["api_key"] }

No MCP Server

No MCP Server Card or live endpoint detected. Model Context Protocol (MCP) is the dominant standard for AI agent integrations, adopted by Claude, Cursor, VS Code, GitHub Copilot, Gemini CLI, and 30+ others. Without it, agents can't take actions in your product.

Fix: Publish a machine-readable MCP Server Card at /.well-known/mcp.json: { "mcpVersion": "1.0", "name": "Your Company MCP Server", "description": "What AI agents can do with your product", "endpoint": "https://yourdomain.com/mcp", "capabilities": ["tools", "resources"], "auth": { "type": "bearer" } } If you don't have an MCP server yet, consider building one , it lets AI agents like Claude take actions in your product directly (create accounts, run searches, fetch data). This is the fastest-growing AI integration standard right now.

No A2A Agent Card

No A2A Agent Card at /.well-known/agent.json. Google's Agent-to-Agent (A2A) protocol enables AI agents to communicate with and delegate tasks to each other , including your product.

Fix: Publish an A2A Agent Card at /.well-known/agent.json: { "name": "Your Company Agent", "description": "What this agent can do", "version": "1.0", "url": "https://yourdomain.com/agent", "capabilities": { "streaming": false, "pushNotifications": false }, "skills": [{ "id": "your-main-skill", "name": "Main capability", "description": "..." }] }