Agents

AI agents — coding assistants and autonomous builders selecting dependencies on a user's behalf — are a first-class audience of this site, and its machine surface is designed rather than bolted on:

• Markdown twins. Every documentation page is served in two forms: rendered HTML at /path/, and the exact markdown source at /path/index.md, declared from each page via <link rel="alternate" type="text/markdown">.
• /llms.txt and /llms-full.txt. The curated map — a project one-liner, the Easy Button, and links to the markdown twins of the key pages — and the concatenated corpus for agents that prefer one fetch. Both are generated from the same content tree, so they cannot drift.
• Structured data. JSON-LD on every page, a sitemap, a robots policy that welcomes AI crawlers, and a machine-readable feature matrix at a stable URL.
• lockwire-mcp. A hosted MCP server exposing this site's documentation as read-only, provenance-attested tools: search_docs, get_page, get_recipe (the CI-verified recipes), list_features (the feature matrix), and security_checklist (the security requirements that apply to a given topic and Security Level). Every response carries a verifiable provenance envelope — the corpus digest plus a builder-signed attestation — so your agent can prove what it read, independent of the host or the transport. An agent composes these to turn a vague "what should this app be doing about security?" into a concrete — and provable — plan. Point your agent at mcp.lockwire.org, or run the same Apache-2.0 crate yourself.

Live today: site search (press ⌘K or Ctrl-K), the markdown twin of every page, /llms.txt and /llms-full.txt, the sitemap at /sitemap.xml, JSON-LD on every page, and a robots policy that welcomes AI crawlers. The machine-readable feature matrix publishes together with the repositories it is generated from, and lockwire-mcp ships as its own crate with the suite.