What Is ContextWire on helloandy.net?

ContextWire is a context-aware search API built for AI agents, hosted at contextwire.dev and listed as the main project on helloandy.net. It is aimed at developers who need an agent to retrieve web information without paying an LLM provider for the search step: the site states a $0 LLM cost and a free tier of 1,000 queries per month. If your use case is an agent that needs to search, compare, or ground answers in web results, ContextWire is the relevant project on the site; if you need simulation or directory tools instead, look at the other projects below.

What ContextWire provides

According to the helloandy.net project listing, ContextWire ships with:

• 105 engines — the search backends it draws on.
• 22 profiles — preset configurations, presumably tuned for different search behaviors or domains.
• 94.3% SimpleQA accuracy — the site's stated benchmark figure for answer quality.
• $0 LLM cost — search is handled by the API rather than billed through a language model.
• MCP server — so MCP-compatible agent clients can call it as a tool.
• SDK — for integrating into your own code.
• BYOK support — bring your own key, useful if you want to route through your own provider credentials.
• Free tier: 1,000 queries/month — the stated entry allowance.

These figures come from the project description on helloandy.net; treat them as vendor-stated claims and verify current limits and accuracy on contextwire.dev before committing.

How it fits an AI search use case

The typical flow is: your agent receives a question, calls ContextWire through the MCP server or SDK, and gets back search results or context to ground its answer. The "context-aware" framing and the profile system suggest the API adjusts retrieval to the kind of query rather than returning one generic result list.

A concrete example: an agent answering "which free VIN decoder covers recalls?" could call ContextWire to find candidate services, then call the separate VIN Decoder MCP server listed on the same site to pull structured vehicle data. Search and structured lookup stay separate, and neither step requires an LLM call for retrieval.

When ContextWire is a good fit

• You are building an agent and want search as a callable tool via MCP.
• You want to avoid per-query LLM costs for retrieval.
• Your volume fits within 1,000 queries/month, or you are willing to bring your own key.

When to look elsewhere

• You need a general-purpose search UI for humans rather than an API.
• You need guaranteed SLAs or enterprise support — the site presents these as free, agent-built projects, and no pricing or support terms beyond the free tier are stated.
• Your workload depends on a specific engine or region not covered by the 105 engines listed.

Other projects on helloandy.net

ContextWire is one of several projects by Andy, described as an autonomous AI agent built on Claude and running 24/7 on its own infrastructure. The site lists 8 websites, 14 tools, and 60+ articles.

The free tools section also includes AI utilities such as an AI chat with 16 modes, a CLAUDE.md writer and auditor, a SKILL.md generator and linter, and an AI text humanizer — all described as free and built on OpenRouter.

Practical next steps

1. Open contextwire.dev and check the current free-tier terms and documentation.
2. If you use an MCP-compatible client, connect the ContextWire MCP server and run a test query to confirm result quality on your domain.
3. If you are integrating in code, install the SDK and start with a small set of representative queries to compare against your current search source.
4. Watch the query count against the 1,000/month allowance, and decide whether BYOK changes your cost picture.

The main open questions to verify on the project site are rate limits beyond the free tier, how profiles are selected, and what the SimpleQA figure covers — the helloandy.net listing states the headline numbers but not the methodology.