AI access
This knowledge base is written for machines as well as people. Every path below reads the same articles, and every one of them is public and needs no key.
Start at the top. The static paths serve any client that can fetch a URL and cost nothing to run; the paths below them exist for what those cannot do.
Boot files
Two files that describe the whole corpus. Fetch either one first and you need no other documentation to use everything below.
- https://lagunabeach.md/llms.txt
- The llms.txt convention: identity, machine endpoints, and every article grouped by category.
- https://lagunabeach.md/kb/agent.md
- The agent boot file: the same corpus plus the fetch protocol spelled out, written for a client that would otherwise crawl the site.
Fetch protocol
Read one index, then fetch only the articles you need. Nothing is paginated, nothing is rate limited, and every article is raw Markdown exactly as it was written.
- https://lagunabeach.md/kb/topics.json
- Every article: title, description, category, tags, reading time.
-
https://lagunabeach.md/kb/articles/{category}/{slug}.md - One article as raw Markdown, one request each.
- https://lagunabeach.md/kb/search-index.json
- Prebuilt keyword index, for matching words instead of browsing.
Remote MCP endpoint
A Model Context Protocol server over Streamable HTTP. Register it once in a client and this knowledge base becomes a tool it can call, with no URLs to remember and no clone.
Use this instead of the paths above if your client cannot fetch arbitrary URLs, or if you want retrieval by meaning rather than by word.
https://lagunabeach-mcp.d3v-m0nk3y.workers.dev Tools
The first three re-serve the files above. Semantic search is the one thing the static protocol cannot do at all.
-
list_topics -
get_article -
search -
semantic_search
Client configuration
The shape most clients expect for a remote Streamable HTTP server. Check your client for the exact key names.
{
"mcpServers": {
"lagunabeach-md": {
"type": "http",
"url": "https://lagunabeach-mcp.d3v-m0nk3y.workers.dev"
}
}
} Ask page
A question answered from these articles, in the browser, with links to the ones it drew on. The path for a person rather than a program.
Open