Agent access
Let an agent query my public portfolio.
This is a dedicated portfolio MCP, not a general website crawler. It is public, read-only and deliberately narrow. It exposes curated site content, not my vault, CRM, credentials, files, or private network.
Try the recruiter lens.
Paste a short role description. This client-side preview uses the same deterministic matching logic available to recruiter agents.
Runs locally against the same curated evidence exposed through the MCP server.
Streamable HTTP
Connect to the live endpoint.
https://mcp.alexeiudall.com/mcp
The endpoint above is the production Streamable HTTP server. It is intended for recruiter agents, role-matching workflows and anyone inspecting the public evidence behind this portfolio.
One source, three interfaces.
The human website, JSON routes and MCP server import the same versioned content package. The website does not call its own MCP to render a page, but the content contract is dogfooded across every interface so claims cannot quietly drift.
The pattern is reusable. This instance remains intentionally tied to my portfolio; a general-purpose website-content MCP would be a separate product.
Tools
search_profileSearch experience, projects and case studies by natural-language query.
match_roleCompare a role description with evidence, strengths and explicit caveats.
get_case_studyReturn one full evidence-backed case study by slug.
list_writingList public essays, topics and source links.
get_contactReturn only the public profile channels shown on this website.
Resources
Stable URIs for agents that prefer resources over tool calls.
portfolio://profile portfolio://experience portfolio://projects portfolio://case-studies portfolio://writing
Agent-readable without MCP.
Lightweight clients and crawlers can use plain text or JSON. These routes are generated from the same content package, so the human and machine versions cannot quietly diverge.