Code to Image API, CLI and MCP

Code to Image is a pixel-perfect HTML/CSS-to-image API built for the AI agent era. The problem: AI agents can write flawless HTML and CSS, but they cannot turn it into deterministic image a human can post, send or print. Existing HTML-to-image services predate the agent ecosystem – they offer neither an MCP server nor a first-class CLI, and the strongest players lock users into template editors. The solution is one rendering engine exposed three ways: a REST API for backend integrations, a CLI for scripts and pipelines, and a native MCP server listed in the official Anthropic MCP Registry – so agents like Claude can generate images as a regular tool call, with no glue code. Output formats: PNG, JPG, WebP, returned as binary or hosted URL. The business layer is deliberately simple: a free tier without watermark, subscriptions from $9/month, and prepaid packs for occasional users, all self-serve via Stripe. The project is also an experiment in solo, AI-assisted product development: one founder took it from idea to live billing in about six weeks, using AI coding agents throughout – a tool for agents, largely built by agents.

MCP-first image API. Solo-built in six weeks, self-funded. Designed for speed and automation. And… the founder is finally eating his own "dog food" ;) What better proof of software that solves a real problem?