Skip to content

feat(site): ready for AI agents - #1

Merged
sanjevirau merged 1 commit into
mainfrom
feat/agent-ready
Oct 23, 2026
Merged

sanjevirau merged 1 commit into
mainfrom
feat/agent-ready

Conversation

@sanjevirau

Copy link
Copy Markdown
Contributor

Makes firenook.dev ready for AI agents, by the checks of Cloudflare's Agent Readiness scanner.

  • Markdown for every page. The site's own pages are converted from their <main> after the build (the docs already had theirs), each page links to its Markdown, and a zone rewrite rule serves it to Accept: text/markdown.
  • A docs MCP server at /docs/mcp with search_docs, read_page and list_pages, on Streamable HTTP. It is the one address that runs code (runWorkerFirst); everything else stays static. The same tools are registered in the browser through WebMCP.
  • /.well-known: API catalog (RFC 9727), MCP server card, AI catalog, skills index and a Firenook skill. /auth.md says nothing needs an account.
  • robots.txt: Content Signals (search, ai-input and ai-train all yes) and the AI crawlers by name. The home page gets a Link header; every response gets Vary: Accept.
  • Docs: /docs/tools/agents.
  • check:site now checks every page's Markdown, every search result's page and heading (431 anchors), and the skill's digest.

Tested locally with cf-wrangler dev: the official MCP SDK client (1.32.1) connects, lists the tools and calls each; WebMCP registers its four tools on site and docs pages.

🤖 Generated with Claude Code

https://claude.ai/code/session_01YXrJymDkQp71Esk3fwMmYV

- Every page as Markdown: the site's own pages are written out from their
  <main> after the build, next to the docs' own; each page links to it, and
  the zone serves it for Accept: text/markdown.
- An MCP server over the docs at /docs/mcp (search_docs, read_page,
  list_pages), the one address that runs code, and the same tools in the
  browser through WebMCP.
- /.well-known: the API catalog, the MCP server card, the AI catalog and the
  skills index with a Firenook skill; /auth.md says nothing needs an account.
- robots.txt with Content Signals and the AI crawlers by name; a Link header
  on the home page; Vary: Accept.
- A docs page, /docs/tools/agents, and check:site checks the Markdown, every
  search result's page and heading, and the skill's digest.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01YXrJymDkQp71Esk3fwMmYV
@sanjevirau
sanjevirau merged commit c087762 into main Oct 23, 2026
1 check passed
@sanjevirau
sanjevirau deleted the feat/agent-ready branch October 23, 2026 16:11
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant