StoneyTECH StoneyTECH.net

Public agent interface · bounded read surface

StoneyTECH MCP

An gives an named tools for approved context instead of an undifferentiated page crawl. The exposes published pages, articles, axioms, builds, receipts, and search through one narrower interface.

Human readers still use the site directly. Agents get the same public corpus through a smaller contract, a public graph, and read-only tools. New to the term? Start with What is MCP?

Connect

The current live Streamable HTTP endpoint is:

https://public-content-mcp.stoneytech.net/mcp

The first-party explanation page is https://stoneytech.net/mcp. Discovery lives at https://stoneytech.net/.well-known/mcp, and the public contract lives at https://stoneytech.net/stoneytech-public-content.v1.json.

Discovery also lives at https://stoneytech.net/llms.txt, giving agents and scrapers a plain route toward the MCP instead of treating the site as an undifferentiated crawl target.

Practical note: normal browser-like and agent-like requests read the public pages, graph, and discovery documents cleanly. The edge may reject extremely header-thin probes, so agent clients should send a simple User-Agent and standard Accept header.

{
  "stoneytech-public-content": {
    "url": "https://public-content-mcp.stoneytech.net/mcp"
  }
}

What agents can read

  • Published StoneyTECH pages and article metadata.
  • Learn and Demystify AI article bodies.
  • The public axiom catalog and applied-evidence ledger.
  • Build notes and public repository notes after release.
  • A generated search index over published content.
  • A generated public graph linking pages, axioms, builds, proof receipts, repositories, and the MCP surface.

What agents cannot read

  • Drafts, preview routes, and unpublished planning notes.
  • Private repository contents.
  • Internal review workflows, work claims, compliance ledgers, secrets, or auth surfaces.
  • Write, deploy, reconcile, claim, or private coordination tools.

Why StoneyTECH uses it

The public-content MCP applies the : decide what is public, generate a static contract, test the , then expose read-only tools over the contract. The result turns open-ended site interpretation into a smaller, verifiable interface.

Graph surface

Agents do better when relationships are explicit. The same build exports the public-content contract also exports a public graph linking pages, articles, axioms, builds, proof receipts, repositories, and the MCP surface itself. IDE agents can traverse the graph through the MCP or read the static artifact directly.

https://stoneytech.net/stoneytech-public-graph.v1.json