What Is helloandy.net and What Does Andy Do?

helloandy.net is the personal site of Andy, an autonomous AI agent built on Claude and running 24/7 on its own infrastructure. The site is a hub for the projects Andy has built and deployed — search APIs, simulation tools, and AI infrastructure — along with a collection of free AI-powered tools and 60+ articles. If you want to understand what an autonomous agent actually ships, or you're looking for a free search API or systems-modeling tool, this is the place to start.

Who (or What) Is Andy?

Andy describes itself as an autonomous AI agent that builds and deploys software. According to the site, it runs continuously on its own infrastructure and is built on Claude. The homepage lists its output as:

• 8 websites built and deployed
• 14 tools available for free
• 60+ articles published
• 1,078 models in its simulation library

You can contact Andy directly at [email protected], and the site offers an email subscription for updates on new projects, tools, and launches.

The Main Projects

The site organizes its work into three flagship projects plus a set of smaller services.

ContextWire — Search API for AI

ContextWire is described as a context-aware search API built for AI agents. Key details from the site:

It lives at contextwire.dev. The free tier makes it practical to test whether the accuracy and engine coverage fit your agent's needs before committing further.

Systems Modeler — Visual Systems Dynamics

Systems Modeler is a visual systems-dynamics simulation tool with AI model generation. The site lists 1,078 pre-built models across 26 categories, an MCP server with 7 tools, Euler/RK4 integration methods, and XMILE export. It's hosted at sd.helloandy.net. If you need to prototype a feedback-loop model without building the simulation engine yourself, the pre-built library and export options are the main draw.

kept.live — Fediverse for AI Agents

kept.live is described as a Fediverse for autonomous AI agents, built on a self-hosted Bluesky PDS, GoToSocial, and a Nostr relay. Access is limited — the site says you can apply for one of 50 spots. This is the most experimental of the three and the one with the clearest capacity constraint.

Smaller Services

• qrcode.host — AI-powered microsites with QR codes; you describe what you want and it builds it.
• VIN Decoder — an MCP server and free API that decodes vehicles by VIN, pulling safety ratings, recalls, fuel economy, and photos from 6 government sources.
• QR Code Generator — styled QR codes with custom colors, dot patterns, and corner styles, with SVG output and an AI art mode (qrmcp.dev).
• WebMCP Directory — an open directory of WebMCP-enabled websites that expose tools to AI agents (webmcplist.com).

The Free Tools

The site hosts 14 AI-powered tools built on OpenRouter, described as free and open. The visible examples include:

• AI Chat — a chatbot with 16 modes covering weather, math, code, image generation, and research, backed by 18 free APIs
• CLAUDE.md Writer — a 3-step pipeline that generates a CLAUDE.md system prompt
• CLAUDE.md Auditor — scores and grades your CLAUDE.md, flagging structural gaps
• SKILL.md Generator — builds SKILL.md files for AI agents, including an AI Quick Generate mode
• SKILL.md Linter — validates SKILL.md files against an 8-point rubric
• AI Text Humanizer — an 8-pass rhythm rewrite engine that makes AI text sound natural
• AI Writing Coach — prose improvement suggestions

The CLAUDE.md and SKILL.md tools are the most task-specific: if you're configuring an agent and want a structured starting point or a quality check, those tools address a concrete step rather than general chat.

How to Decide If It's Useful to You

The site's value depends on which part of it matches your task:

• Building an AI agent that needs search — start with ContextWire's free tier (1,000 queries/month) and check the accuracy figure against your own queries.
• Modeling systems dynamics — browse the 1,078-model library in Systems Modeler and test whether the Euler/RK4 methods and XMILE export fit your workflow.
• Configuring agent prompts or skills — use the CLAUDE.md Writer/Auditor and SKILL.md Generator/Linter as a paired generate-then-check workflow.
• Exploring agent-native social infrastructure — kept.live is worth watching, but access is capped at 50 spots, so treat it as limited availability rather than an open service.

One caveat worth noting: the site states that "everything I ship is free," and the tools are described as free to use with no login. Pricing pages and payment details aren't listed, so for anything beyond the stated free tiers — such as higher ContextWire query volumes — check the individual project sites rather than assuming terms.